I briefly reviewed this, and it appears to be a valid problem. It's not new to FCP 11.2 but has existed in some form since back in 10.8.1. I only had time to test it on Sequoia 15.7 and Tahoe 26.0.1, but I vaguely recall seeing this earlier but didn't understand it.
In brief, it appears like a "color space state problem" whereby FCP loses track of the expected color space behavior of a proxy file generated on a ProRes RAW clip. Sometimes turning on and off a built-in camera LUT or skimming the browser clip will cause the color behavior of the timeline clip to change. That makes it hard to test since aspects of the behavior are a bit unpredictable.
Testing is further complicated by the numerous combinations, such as wide-gamut library, standard-gamut library, Rec.709, HLG and PQ projects, whether using a built-in FCP camera LUT, etc. The color conform settings also would need testing for all those combinations.
I think the proxy behavior is supposed to track the ProRes RAW config at the time proxies are generated. It won't let you change the RAW settings after they are generated. Another possibility is the program logic that tracks that could be getting confused. Also there are limitations on H.264 proxies vs ProRes Proxies.
Maybe the simplest way is importing a ProRes RAW clip of a colorchecker card to an SDR library, then change RAW settings to use a log format such as S-Log3. That is a very common usage mode.
Then add the clip to a Rec.709 timeline and observe the luma waveform of the colorchecker card. Then create ProRes proxies on the clip, switch back and forth between proxy and original mode. It should not change. Then turn on/off the built-in S-Log3 camera LUT and see if the waveform and viewer persistently change after the first on/off cycle. It should not -- it should always go back to the prior state.
You will often see the waveform y-axis scale change between IRE and a luma scale, which implies it's internally switching between viewing that as an SDR vs HDR clip -- even though both library and timeline are SDR. But it does similar odd things in a wide-gamut library and if using a Rec.2020 PQ or HLG project.
It's not 100% certain it's an FCP problem. It could be a MacOS framework layer such as ColorSync. I think some DaVinci Resolve editors have reported similar behavior, now that Resolve 20.2 supports ProRes RAW.
