Inconsistent Generic RCS Configs

[Capkirk123]

Generic RCS configs are, well, a little inconsistent with their performance, both with regards to ISP and O/F ratios.

With regards to ISP, I did some calculations in RPA (using a 15AR, 0.75 MPa Pc engine as a baseline) and discovered that while most RCS configs have ISPs about 11-12% lower than the numbers predicted by RPA, there are some notable exceptions, such as HTP performing 16% worse, while Hydrazine only performs 2% worse Upon further observation, RPA consistently underestimates Hydrazine performance. HTP is a bit more complex, as HTP runs cold enough that water vapor crystallizes in the nozzle at low expansion ratios, resulting in a hard upper limit on performance. I assume these performance numbers were based on real engines, and minor variations in the design of those engines resulted in the variations in performance seen as compared to the theoretical performance.

With regards to O/F ratio, the O/F ratios chosen for RCS thrusters seem a little odd. The thrusters using MMH have O/Fs around 1.65, clearly based on the R-4D and AJ10-190. The Aerozine config uses an O/F of 1.625, which matches no engine I can find, but is still pretty close to the previous 1.65 value. However, the UDMH config has an O/F of 2.6, which is significantly different from the other configs, and matches no engine I could find (Soviet pressure-fed UDMH/NTO engines almost universally operated at an O/F of 1.85).

Finally, there is the issue with tech levels, which I believe has been discussed before. The only way for RCS to achieve the performance given by higher tech levels is by significantly increasing expansion ratio, but increasing the tech level does not change the size of RCS or generic thruster nozzles. Further, increasing the tech level increases both SL and Vac ISP, when in reality increasing expansion ratio would be expected to decrease SL ISP while increasing Vac ISP.

The first issue should be relatively easily resolved by just setting RCS ISP modifiers based on calculated theoretical values, rather than trying to extract trendlines from IRL systems, although this will be minorly save-breaking.

The second issue is a bit more complex. The unusual O/F of the UDMH+NTO config can be easily resolved, but will be savebreaking. It would also be possible to address the varied O/F ratios of real systems with more configs (potentially through adding subconfigs to generic RCS), but this would drastically increase the already large selection of RCS configs for relatively minor gain.

The third issue is a bit more complex, with potential solutions ranging from rescaling RCS thrusters with Tech Level, to moving away from the generic RCS system completely in favor of real engines. This probably deserves an issue of it's own.