Man, this one often starts a debate. And I don't have time right now for a long reply. But a couple of thoughts for you to ponder till later:
There's lots of ways to do this, and while important, its not worth the obsession some put into it.
The first thing to address is recording level. I find it nearly universal that people record at much hotter than ideal levels.
Next is the effects/insert chain, you want to be aware of a good level going in, and coming out of any effects you use. There is no hard fast rule. Some plug ins are affected by level far more than others. If you don't use something intending to add gain, you should look for a similar input and output, as a general rule. It doesn't have to be micrometer accurate, using your ears and disabling/enabling each plug-in is generally sufficient for this.
FWIW, I've found SOME guitar amp sims seem to respond well to a hotter signal than I would use elsewhere. Compressed before the sim sometimes too.
Track level (the last question)-
I've said before that I would discourage using that project as a gain staging example. I cant speak for the producer of that piece, (Mitchell) I can only guess it was either intentional to get a certain sound, or it was something quickly put together to use for purposes that didn't include gain structure.
Whichever it is, my comments would be on one hand- within the DAW, those "in the red meters" aren't hurting the sound due to 32bit floating integer processing.
Short explanation- cant clip internally due to that alone.
In general your end concern is rendering and what you are going to present to your D/A converter. And your intent. IE: Are you mixing or mastering?
For mixing purposes, a good general "rule of thumb" target is to leave some headroom. A minimum of -6dBfs is a common recommendation on the master bus.
And obviously, an accumulation of tracks hitting the red, or even close isn't going to sum to -6, so....
This is where it gets complicated, and where even more diverse methods tend to be shown.
I can only explain how I do it, which is pretty conservative due to my preference for dynamic and "open" music. And that is prone to tweaks and changes as I learn and experiment.
But high RMS and hyper compressed "modern" mixing just isn't something I really appreciate.