Yes, I'm quite familiar with how compression gets used in production. I'm also aware that, of all of the types of processing I've used over the years, compression is the one you shouldn't print to multitrack with. When doing your initial gainstaging, you need ALL of the tracks to have a similar dynamic going into the mixing/rendering stage so that it makes it easier to determine the general levels of elements within the mix. At that point is where you'd want to compress...on a strip insert, going into the mixbus. If you've compressed to the multitrack, you then have no dynamic leeway and the resultant track(s) won't "breathe".
About the only times I've ever compressed to multitrack were on randomly-peaky sources, mainly vocals. And invariably, I was pushed into that by the vocalist's incompetence with performing for a mic. Vocals that swing between -20 and blasto dB really don't make for fun mixing work.
If you're trying to do "loudness wars"-type tracks, then I could see the point of repetitively mashing, mashing, mashing the dynamic range. But keep in mind, also, that people are REALLY tired of that squashed sound. It might make radio happy, because they always shoot for having the "loudest signal" in any market, but it winds up sounding mashed and crummy...mainly because, like in the first paragraph, the music isn't "breathing".