Living on the surface of Mars in great amounts is impossible in any case, due to the low gravity, low magnetosphere, low insolation, and so on.
They will preven any healthy children development.
While the magnetic field and the insolation can be artificial in small amounts inside the habitats, the 0.4 g is a no-go even for long-term adult living, let alone the child growth.
The orbital habitat with artificial gravity is the only way to live in/on.
It IS their biggest problem.
The O'Neill cylinder is a harmful, purely pseudoscientific fantasy, having nothing common with dull reality and elementary physics.
They should be blamed and degraded like a kind of perpetual motion machine.
Humans need food.
The food is O,C,N,H combined into molecules. This needs energy. The energy comes either from the Sun, or from an artificial energy source like a nuclear reactor.
The process of the food molecule appearing is highly ineffective in sense of energy, and this can't be changed for purely thermodynamic reasons,
because you can't change the total enthalpy and entropy of the chemical reactions turning CO2, N2, and H2O into polysaccharides, fats, and proteins.
Using reactors instead of Sun also changes nothing in this sense.
So, the energy conversion efficiency of the chain will stay same low as now, and the "entropical" waste heat will always prevail over the useful energy stored inside the food molecules in form of atomic bonds.
But even the food-stored energy gets released when a human digests the food, because it's the main purpose of the food.
Most of the digest-released energy dissipates as waste heat when an ATP molecule splits to ADP, then to AMP one.
Some small part of energy gets stored as the human body molecular bonds, but gets released later, when the body tissue in some form decays.
So, all energy, coming from the energy source nto the food production, finally (and mostly immediately) gets turned into waste heat, requiring a heat sink.
The whole food chain and life support is a thin gasket between the heater and the cooler, as any other useful thermodynamic process.
Here, on the Earth, the cooler is the huge planetary body, and especially the large amount of cool water in the ocean.
So, the people remember only about the heater, which is usually the Sun.
But rarely they have to care about the cooler, because their amount and total power is too negligible to warm the oceanic cooler. Only last few decades they start thinking about this at all.
The waste heat washed from the human farms and habitats into the ocean, gets radiated into space as IR photons.
Because any kind of orbital habitat, including such large one as a planet, has to release the same amount of energy as it had gotten from the energy source, and the only way to do this in vacuum is radiation.
So, any orbital habitat needs as much energy income per human as it has on the Earth, and enough large radiator panels to radiate this total power into space.
If it works on the Sun energy, it needs as much solar income area as on the Earth, and the plants should be splattered all over this insolated area as a thin layer, because the light can't infiltrate deep.
Any light passing below the leaves (or algae) is not caught by them.
The Earth radius is 637 km, but the useful humus layer is just about a meter thick, and the biosphere if press it down is just several inches of organics more.
Becausse the sunlight can't go deeper, and the chemical reactions are thinly splattered over the plowland/greenhouse area.
On the Earth a humanrequires about 1 hectare (10 000 m2) to fulfil all his needs (food, clothes, basic medicines, etc) and 0.1 hectare to barely survive.
A Sun-powered orbital (or cloud balloon) habitat needs same. Maybe even several times less if produce fertilizers and optimize the plant structure, but this doesn't change the game.
This means than either the orbital habitat should be a enormously huge thin plate(thus forget about rotation, but face the problem of coolant pumping through the enormously long pipes),
or it should be a compact cylinder (like the O'Neill's one) of ridiculous cross-section area. And the large is the cylinder, the lower is the insolated/living area ratio.
As you can easily calculate, the classic ~10 km sized O'Neill cylinder has the insolated area to support just several tens (optimistically - hundreds) thousands of people. This certainly looks absolutely silly to seriously plan.
They may decide to use fusion reactors instead of the sun light, and increase the insolated area by using pipes with algae around light bulbs.
This indeed a soltion to the problem of the insolated plowland area, and can pack many millions of people inside the cylinder.
But this still needs a heat sink, and the amount of energy to be sunk stays same.
When you use the sun-powered food chain, the productive layer is so thin that the light receiver itself (plowland+water pool) is enough large to be a radiator panel.
In the reactor-powered habitat they need to have radiator panels of similar total area like they need for the plowland.
So, the problem stays same, just instead of enormous flat solar receivers they still need enormous flat radiator panels of comparable total area.
This in turn again forces to limit the total power of the internal power source with same tens-thousands of people per cylinder.
So, the only way to colonize anything beyond the Earth is to separate the habitat and the food production completely.
The habitat can be only orbital, rotating at 1 g.
The food production doesn't require anything of that if use algae.
The algae are always in zero-g due to timy size and hydrostatic balance, they hardly need magnetic fields for same reason, they can use a pool of any shape to grow, either a thin layer of glass pipes under sun, or the same glass pipes around a light bulb.
But the waste heat still needs to be utilized. So, orbital greenhouses for millions of people should face the sane problems of cooling and coolant pumping like in O'Neill case.
All you gain is absence of rotation, so your orbiatl plowland won't be torn aparn by centrifugal force.
So, the only thermodynamically reasonable way of the agricultural mass production is a thin layer of sunlighted greenhouses covering a planetary surface.
In this case they are cooled same way like on the Earth, by the panetary body to redistribute the heat, and the planet surface are to receive enough power from Sun and the radiate it then as waste heat.
You can locally enforce this by adding reactors, but their total power shouldn't be comparable to the solar income, as in such case they will overheat the system over the plants critical temperature and kill the agriculture at all.
***
Thus, the only viable way to colonize some place in the universe is to have a massive planetary body inside the habitation zone of the star, cover it with thin layer of algae-pipe greenhouses, and make this the source of food for you colony.
The colony itself should be only orbital, in rotating habitats (though not the huge cylinders, which is a loss of volume, but bunches of smaller cylnders attached to a zero-g central module).
The population of the orbital colony is limited by the food production of the planetary surface, and as unlikely it can be comparable to the Earth one, the colony capacity can be only much lower than the Earth population.
In the Solar System the only colonizable place is Mars.
The Moon is too small and too close to the Earth to bother.
The Ceres is too small, receives much less solar energy than even Mars.
The Venus could be potentially turned into a greenhouse, if remove the atmosphere and let the light pass to the ground, but it doesn't rotate, so the on-ground greenhouses would face monstrous problems with cooling in the eternal midday, as they had to pump hot water to another part of the planet.
Cloud balloon greenhouses are also not a variant, because they require same total are, and thusabsolutely enormous size of balloons making them fully unstable mechanically, in the windy upper atmosphere.
So, in any case the whole Venus could produce much less food than it may seem based on its size and gravity.
Actually, a totally useless place in sense of colonization. Maybe some minerals can be mined, but also unlikely due to lack of long-term liquid water in its visible history.
***
So, the Mars is the only place to be colonized, but its maximum food productivity is many times lower than the Earth have, due to much smaller size and twice lower insolation.
It can support a hundred million or several times more, but this is not what makes it reasonable for demographic migration at all.
Thus there is no reason to keep hundreds of millions there,
It can be just a backup multimillion-capable colony, populated by small amount of scientific and industrial personnel.
***
The interstellar colonization is absolutely the same.
For the reasons above they can widely colonize only a water-rich planet in habitation zone, with small amount of people in orbital habitats.
It makes sense to colonize only a star system of not less than ~0.5 bln capability, otherwise they can't be self-sustauned. Seed or not, doesn't play a role.
The natural biosphere is totally based on the sunlight, so it's just a less productive version of the planet fully covered with algae greenhouses.
