For radio broadcasts I use the compressor on the main output with a threshold just below clipping level and a high compression ratio. And, of course, use the faders so that the compressor threshold is never reached unless something unexpected happens.
I use the on-board limiter on the Main L/R in RMS mode and in parallel compression. Since PC does not protect from OL, I need to be a bit conservative in my output level but I think that could be fixed with a “brick wall” limiter as the last thing in the chain. I think the PC on the SQ sounds mighty good. YMMV.
In my albeit limited experience this kind of requirement is best met by multistage compression/limiting.
So put a compressor on the mix bus but then maybe pick up the stream output on a matrix send and put another compressor on the matrix set to infinite ratio, ie a limiter. As to which of the available compressor models to use, I can’t really comment on that.
I’ve set up multiple groups and a matrix for our live stream to do just this. I find it works pretty well and allows compression to be added in stages exactly as others have suggested. I just use the built-in compressors and am happy with the results. I’m not near the desk to give specifics but I did a reasonable walk-through of the setup for our volunteers which you might find helpful: https://youtu.be/WsTh9L3-vZ0