Stability derivatives may fluctuate slightly when a craft is reloaded in the editor.

[Rodhern]

I noticed that even in my newest (experimental) commit I get small fluctuations in the calculated results. If for instance I load a craft and calculate the Cl and Cd, and then do it again, then I get little almost unnoticeable differences.

The differences I have seen so far (fingers crossed) are nowhere near the ones we experienced a while back with the parts cache issue (ferram4#177). Even so, it seems a bit mysterious to me.

[Rodhern]

I think I can reduce the variations if I update this signature:
private double CriticalMachFactorForUnsmoothCrossSection(VoxelCrossSection[] crossSections, double finenessRatio, double sectionThickness)
to this:
private double CriticalMachFactorForUnsmoothCrossSection(VoxelCrossSection[] crossSections, int frontIndex, int backIndex, double finenessRatio, double sectionThickness)

It does not eliminate the fluctuations but they seem to become less pronounced in my example(s).

[Rodhern]

If you still notice these fluctuations in the newest release, tagged v0.15.9.1_Rodhern_update_4, and particularly if you are able to tell under which circumstances it happens, please let me know.

[Rodhern]

Notice: I have confused Github by merging a local dkavolis branch copy. The "dkavolis#3" issue that dkavolis reference in d9665fb is confused with "#3".