Er, that 10 dB I mentioned IS the redirected bass. The 10 dB implied gain on the sub channel is, well, implied. It does not take up dynamic range.You're forgetting about the redirected bass from other channels. In the case of 7.1 you end up with 20.2dB required headroom. Then add 10dB for speaker trims and 10dB for EQ and your headroom requirement becomes >40dB.
Now if you send that data via the limited capabilities of HDMI you're making your system worse and not better. Better would be to keep the 32bit audio data right up until the DAC.
Speaker trims and EQ will eat some too, but thst's an argument for (a) well-matched speakers and (b) room treatment before EQ.
But it seems the crux of your argument is maintaining 32 bit resolution of signals with gain instead of 24 bits. This does not help.
You see, at best you have 24 bits of resolution to begin with. That is not bad:144 dB dynamic range, so losing 20 or even 30 dB is fine with a 115 dB SINAD. All you are doing in preserving 32 bits is raising the noise in your system. If you "overflow" in a processing stage and need 25, or 26, or 30, or 32 bits, you just divide the signal back down to stay in range. It's the same with digital volume control: bad with 16 bit data, OK wirh 24. Remember: you only need enough to match the SINAD of your worst stage. 115 dB SINAD means anything more than 19 bits resolution will be lost anyway.