I am not sure that I see the distinction between "pixel processing" and "parametric processing".
I think this is a crucial distinction and if you don't understand, then that's an issue. To be clear, everything ACR/LR does is only an adjustment of the parameters for a particular control–you can go back into that control and alter it at any time. And, also important to understand, the order the user works in is irrelevant to the order that the parameters are processed.
Compare that to pixel edits where editing the pixels result in a different result after the correction. It implies edits are destructive (and they are unless you use Adjustment Layers). The more you edit pixels the more they fall apart because, well rounding errors and reductions in precision tend to accumulate...
So, bottom line: you want to stay parametric right up to the point where parametric breaks down because of too much parametric data and move into a pixel based edit–where more edits will generally result in less optimal results.
You must grok this...right? A raw file's scene refereed data .xmp settings can be changed ad infinitum (meaning forever) but an output scene referred file degenerates with edits & iterations...
Look at this way...for raw files, as time goes bye, raw processing seems to get better. From the standpoint of rendered image files, the more you edit, the less precise you images become. Will it obliterate you images making them useless? Nope, but is asked, I would say it steps on your image quality potential.