Would you have been willing to tolerate such meddling?
Of course. I am very much non-precious when it comes to what happens to my deliveries down-stream. It's all just raw material for the producers, director, and mixer to mold and shape as they need to get the results they want. Being precious about whether your rear channels are loud enough on the stage is about the same as being precious about whether the dialog and fx are drowning out the score. One or the other is bound to happen pretty often.
I usually tell the mixers / producers / director some variation of this: "I don't care if you discard three of the seven stems, pitch-shift one of them down an octave, and play the other two in reverse. Good luck, have fun, let me know if you need anything more." This usually results in a sigh of relief from mixers who have dealt with too many too-precious composers.
Knowing that liberties are likely to be taken, I don't put anything in the rears that I can't live without, so if they are too quiet or just not there, no harm no foul. And I give detailed notes like, "This cue has the jump scares swooping from the rears to the fronts, and if you just play the fronts then it may sound like they're too short and starting too late, so if you're not using the rears then maybe fold the rears into the fronts if it's sounding weird." - or - "This cue has an extra-unsettling array of those atonal string effects quadruple-tracked in surround, so if there's one spot in the whole score where the rears should be good and loud, this is it."
Another thing that I do which is maybe not typical (or even advisable!) is NOT to allocate elements into the stems by any hard and fast rules like, brass low, brass high, etc. This is partly because I'm almost never working with a conventional layout of orchestral elements - it's more likely to be stuff like: StemA = two sub-bass booms and low hits on jump scares, StemB = just a synth pulse, StemC = evil low synth drones, StemD = distorted midrange textures, StemE = guitar feedback echoes, StemF = icy high strings sustains and high string shrieks on jump scares, StemG = that weird dissonant choir at the start and the dissonant woodwinds in the middle.
So I try to make each stem as close to a complete a "piece of music" as possible, while still trying to have as little overlapping of elements within each stem, and grouping things in sonic categories. So, thin-high-sustainy stuff doesn't go on the same stem as sub-bass booms, but if there's room to split the layers of that thin-high-sustainy stuff across 2 or 3 stems that are otherwise empty at that spot, then I will. If all the high-long strings are on the same stem then they can't do anything with it except turn it up or down. But if the atonal layer is on StemE and the sul-pont sustains are on StemF and the high synth that doubles it is on StemG then they can sculpt things more easily if needed.
This approach might not work for folks who are doing more conventional orchestral layouts, but for my weird smorgasbord of hybrid elements it works pretty well. Because the director usually isn't going to say, "Let's lower the sul-pont strings and raise the behind-the-bridge tremolos", they're going to say, "Can we turn up all that high chaotic screeching? I love that stuff." So for me, obeying conventional "rules" about what sounds go on which stems would be more limiting, both for me and for them. If I did it that way, most of the stems would be empty a lot of the time, with too many elements crammed onto the few stems that they were "supposed to" be on.
Anyway, if my deliveries solve the problem as-is, then... great. But if they don't, what am I gonna do? Sit there on the stage and complain that the explosions are drowning out my precious symphony? That's no way to make friends. And I've had plenty of situations where the music editor has taken snippets or zingers from one cue and laid them across a different cue to solve some problem. Huge relief to have them solve the problem on the stage, on the day, rather than have me scramble to slap a band-aid on the thing. On my last project the music editor took 30 seconds of a pulse+drums bed from one cue, and used good old Serato Pitch-N-Time to speed it up from 110bpm to 122pbm and pitch it down three semitones. Sounded fine, problem solved, note addressed, scratch that one off the list.
Of course, on some tv series I've done I was just delivering stereo mixes, so if it didn't work as-is then the only solution was to go looking for an ALT of the cue. But when I deliver stems I let 'em go wild on the stage if they want. Fine by me.
