Ok I have spent time experimenting on this ...
Take tone and on clip level pull it up high say to around -4db level. Add soft limiter to pull it down to say -12db ...put that clip into a container and play it ... the soft limiter effect seems to be ignored
However, the multiband compressor does appear to work, and the limiter effect maybe works a bit. As in the result in my logic should be identical ... if everything is clip based then surely how that clip playbacks on the top of a timeline or a timeline inside a compound clip should be identical. What's even more odd is that the results change depending on the audio level... so if you repeat the same tests but at lower levels..maybe clipping a test tone to -18db then it holds true the clip itself and the compound clip play the same. The same problem happens if you 'nest by taking a timeline into a second timeline. This is not a V20 issue it has been around since I can't remember.
Maybe I don't properly understand the internal architecture, as I can't believe that this has not been a big issue for people using fairlight... if you containerise an audio mix you just can't be sure that the result remains true. Maybe this is what Avid would call a 'feature' when most of us would say buggered.
My only workaround is to do a mixdown each time. It would be great if BMD could fix it