A description of how I go about the production process - not unique in any way and how most involved in production would go about it (albeit with probably different workflow detail):

a) Open a new Project (from a template file) and name it. This will already have Group channels in it: Vocals, BV, Guitars, Keys, Drums, Bass, etc. Also FX channels – reverb, echo, and Output channels.
b) Import the raw files as available into separate tracks.
c) Arrange the imported tracks in an order, rename the tracks if required. (Check if mono or stereo and fix accordingly.)
d) Route the imported tracks e.g. any vocal tracks would route to the vocal group channel, any guitar tracks route to guitar group channel, and so on.
e) Run the piece, very rough outline mix – set approximate levels, very basic eq at this stage. Just gives a feel for the track as well.
f) Run again to identify initial “repair” work – background noises, hisses, hums, clicks, bumps, etc, on any track.
g) Carry out basic repair work – removal of unwanted noises, etc. This is done on an individual track basis, sometime listening to it solo. Can involve EQ, automation, possible direct waveform editing, etc. This is really just an initial run – will be revisited again a number of times much later in the production process depending on the intrusiveness of the issue.
h) Decide if certain tracks are only “live” at various points and can be muted or volume level dropped using automation between these points. (This process can drop background noise levels and clean up a track especially during quieter parts.)

Once the above (the “craft” side of the process) has been completed (for this stage!) the actual production can be started:

i) Listen again and reset initial volume levels.
j) Balance and outline mix some individual Groups: BV, Guitars, Keys, Bass, Drums (drums in themselves – up to in some cases 12 or more channels - can take an entire session to get close).
k) Run piece again this time with attention to the individual tracks from a content point of view:
Vocals: attention to especially to EQ and dynamics – detailed EQ is vital and also a decision on what level of compression is necessary – is it mainly to catch occasional peaks or is more general compression required? (Sometimes two compressors with different characteristics are run in series to catch different areas.) Identify any plosives or sibilance needing removed (either manually or using a plugin). Is any volume automation required? (vocalists are renowned for varying volume levels due to a number of factors including just varying distance from or angle to the microphone over the course of a piece). Is any EQ automation required on vocals? – similar reasons to volume variation – should manual automation be required for this or can a dynamic EQ be used (or a combination).
l) The above process for vocals needs to be applied to all the tracks to some extent or other.
m) This stage is probably the defining stage for the track: Set a soundstage – this can be defined in terms of the creation of a “space” – partly ambient based and the positioning of instruments within the space. (Often vocals at the front, drums close and central but with some width, BV and certain synths way back and quite wide, guitars closer but spaced, lead guitar and maybe piano off centre on each side.) Actually no rules for any of this – just need to go for what sounds right!
n) Decide on any “build” aspects – should e.g. choruses have different balance, compression, volume, should any saturation be applied? Again no rules.
o) Additional fx – ambience, build – echo, delay.
p) Repeat i) to o) multiple times over a time period.
q) Sort out exact start and end points for the track – should fades be used?
r) Export trial mixes and stems as required throughout the process.

The production side can take a while to get closer to the desired result. Multiple revisits, listening on different systems – earbuds, car systems, Bluetooth blobs, etc, then minor incremental re-tweaks. Also gaps between sessions are required to get perspective – days or longer maybe. Direct feedback from listeners – it is possible to get too close to a track and sometimes areas become obvious when someone draws attention to it and you then hear it as they are hearing it.

The production stage requires a “vision” of the piece – this maybe comes both (I would argue) from experience and also knowledge of the individual musicians – their expectations (and also their capabilities). Without getting too melodramatic you have to “fall in love” with the piece (at least for the period you’re working on it). (You’ll end up listening to it a hundred times or more – certainly I suspect more that the musicians will listen to it! You become aware of every single note.)

Stems: these are generally direct Group outputs (rather than individual tracks) – Vocals, BV, Guitars, Keys, Bass, Drums. Their use (for us) is to give contributors a simple monitoring environment (and also one sort of going a mix direction). Works well with Bandlab (but can equally be used with Reaper or any preferred DAW). E.g. we can have one stereo stem for drums– this stem has all premixing, repairs, spacing, EQ, compression, ambience, applied to it and can be raised or lowered by the contributor to suit their requirement for monitoring. (You are correct – stems are for monitoring only – can’t be used for editing.)

No issue with a master “repository” of original raw takes being kept in Reaper (or even just as wav files). I actually keep copies of all contributors’ original raw takes (in their format as delivered).

Mike