The main thing that can be venting in NASSP is waste water. Your tank is probably full and you have it set to automatic relieve. That would be the pressure relief selector valve on panel 352. 2 is the automatic setting, Dump A and B are manually. Maybe set it to Dump A for like 15 minutes and then switch it back to 2.
Also, just to make sure because you had performance issues, you are using the Orbiter_ng.exe and not the Orbiter.exe, right? Orbiter_ng is required for the DX9 client. Orbiter.exe uses the default DX7 graphics and I remember it having some problems with particle effects (like venting) being very exaggerated. Aside from it performing much worse than DX9.
Your P21 result is a bit worrying, it does use an integrated trajectory, so it should show accurate numbers. But I wonder if that isn't more of a problem with your state vector than your actual trajectory. How much DV was MCC-4? Normally that would be a smaller burn, below 5 ft/s. Maybe you have an alignment problem or something. You did the P52 option 1 before MCC-4, right?
And about the residuals, one pulse from the accelerometers to the CMC is about 0.19 ft/s. So it can't actually resolve residuals more accurately than that. So trying to get any residual of 0.1 ft/s down to 0 is quite pointless, you will just end up with -0.1. I would say 0.2 is the limit of what I would try to trim, usually I don't bother until 0.3.