Damage internal flux fixes

[NathanKell]

Three fixes to the extra flux applied when a part is damaged:

• Use postShock external temp in all cases, because that is the actual temperature outside the part (vessel.externalTemperature is the pre-shock temperature, which is not the actual temperature the part is subject to except in rare cases--and in those cases postShock temp is equivalent).
• Don't double-count temperature by using flux, instead use the coefficient. In particular, heat transfer is proportional to the coefficient times the temperature delta (by definition of the coefficient); using flux but scaling it by a function of temperature uses temperature twice.
• Lerp between newtonian and mach convection in the same way FlightIntegrator does, rather than having a sharp break at Mach 1. This preserves consistency with the rest of KSP's convective thermo model, and respects whatever PhysicsGlobals settings a KSP install has.

[Starwaster]

I have a small concern about finalCoeff in that it has heatConvectiveConstant which I don't think should affect what flux leaks into the interior through the hole burned into the ship.

But not too many people use it and it's probably not that big a deal...

[NathanKell]

Yeah, it's not unreasonable to divide by it and then multiply by the default value here (well, check for it != 0 to avoid a divzero).

[NathanKell]

Heh, that said this code is no worse in that regard than the original, no? Since the original used the actual flux, which is post multiplication of heatConvectiveConstant anyway.