Ext3h

That's a wing, not a stabilizer. It applies lift orthogonal in one direction (asymetric profile). Displayed is the lift the surface would produce at 0° angle of attack. So some lift pointing sideways is to expected if you place just a single wing without radial symmetry. The curious part is that the lift from the main wings has gone missing in the preview, that lift vector is now only from your rudder. The actual issue appears to be that lift for wings is displayed incorrectly at seemingly random for axis-aligned wings - it's off 180° when that happens, potentially negating the lift vectors from symmetric parts. If you tilt the entire craft just slightly off-axis, all wings which are no longer axis-aligned now have the lift previewed correctly again. Easy way to reproduce this? Take just a single wing and mount it horizontally, then rotate it up-side down. Increase the thickness of the wing so you can clearly see the profile, to make sure you got it right (curved side should be on the bottom now!). Now as long as the wing is exactly horizontal (even though already upside down!), it will preview lift as if the lift vector was pointing straight up. Tilt the wing just by 5° in either direction, and suddenly the lift vector flips upside down as it should. If you add more wings, you will then find that it's the individual wings components vectors which have their signs incorrectly flipped.

Isn't the engines exhaust even causing additional convection, eventually even cooling the "object to be heated"? Or is that just the FX implying that? As for the heat capacities, I'm now somewhat missing an intermediate state between "nothing happens" and "fireworks". Something as simple as structural weakening. Reducing part strength noticeably as the heat level approaches max temperature, followed by hard function fail for all active functions (including live support....), and only finally explosion. Getting structural parts to actually "blow up" doesn't even sound right to start with. You would need a huge plasma torch or alike to heat them far enough for that to happen. Breaking apart due to structural weakening and physical force applied sounds far more reasonable.

Looks like you've hit terminal velocity much earlier than you were used too, thanks to reduced drag, and thereby running into the Mach barrier which would make you waste LF/O during early stages without any dV gain. Just try to avoid convection FX at all cost during ascend, and you will notice how your fuel efficiency increases big time.