I noted that most of the quotes Kevin listed come from the recording world. Where plug in processors can "look ahead" and latency is basically the amount of offset cause by the most power hungry or combination of plug ins that the system offsets itself with to keep everything sample aligned.
Digital summing got a black eye as there was noticeable sound degradation the more tracks were being mixed together. This engenders tons of outboard summing boxes where each channel would be run though a D/A summed with an analog mixer or even a simple resistive summing matrix, and back though an A/D into the 2 bus of the DAW. The absence of digital summing was deemed better than the two extra conversions even though folks recognized that as less than optimal. Eventually it was noticed that the digital summing wasn't so bad if the tracks were -12 to -18dB instead of nearly FSD (a legacy analog technique of maximizing s/n by tracking everything as near [or even over] clipping). So now many folks have gone back to "in the box" mixing. That sounds similar to me to the A&H concept of extra digital headroom. And I think many digital live sound mixers are deliberately staying below FSD as people are still used to tickling the reds on analog mixers. I've seen it opined here that some digital desks with "better" pres are basically configured with a bit more headroom and some sort of graceful clipping that tolerates the old analog techniques.
This is an "old wives tale". Digital summation by its very nature is "perfect". That's not to say that other factors may be at play, but summing boxes fall into the same category as external wordclock improving jitter, it's urban myth at best, in reality, hogwash. Digital is capable of far more pristine sound than analogue, period.