Yes, I did generalize for a tapped horn, and now that I think about it, I might not be correct even for that case.
Several 10s of milliseconds sounds very high for "room acoustics" effects. A full cycle at 60 Hz is 17 ms. If you are delaying more than that (in addition to distance and "internal" effects), then you are probably adding unnecessary group delay, which likely impacts transient response sound quality and slam.
FWIW, I've been studying this problem quite intently lately, trying to improve integration between my speakers and subs. Unlike most people, I have practically unlimited DSP resources to throw at the problem, where the only real practical limit is latency. I would say that the processing capabilities built into AVRs and most processors are woefully inadequate for achieving an optimal outcome. The"THX "LR4 sub/sat crossover" is largely fantasy that rarely occurs in real world conditions. The best that most people can do is a brute force evaluation of different XO frequencies and sub delays, where typically response is only optimized on one channel and at one seat. Yet even this effort requires more sophistication than most users are capable of. (Readers of DataBass and some of those who read AVSForum are obvious exceptions.)
No wonder a lot of people prefer bass from 2 channel full-range speakers vs. subs. While the in-room "placement" of the LF drivers in such speakers is non-optimal, the XO is (ideally) optimal for that placement. I've noticed that good anechoic-flat full-range speakers, when pulled far enough from walls, can deliver impressive slam; whereas many sub systems including many with big horns or many large drivers struggle in this respect. My recent experience suggests that phase (or rather group delay) effects are more important than most people realize. And it's not what people think. I.e., a ported sub isn't necessarily sloppier than a sealed sub, though that obviously depends on the competence of design. Such effects are largely minimum phase. (A good thing.) Rather, it is the excess group delay, which arises from crossovers and distance differences that appears to be important. Pre-ringing in particular seems to really kill tactile slam, and it should be noted that FIR filters are not the only way to introduce pre-ringing into a system. Pre-ringing can arise merely from placements and/or delay settings. Any situation in which sound from a sub may reach the listener before sound from a speaker potentially involves pre-ringing. Rooms with rear subs are likely to exhibit pre-ringing for rows behind the one used for calibration. What's not at all clear is where the perceptual thresholds lie for hearing and feeling of pre-ringing effects.
Anyway, I still have a lot of work to do here, and at some point, I may try to do some more formal testing of excess group delay effects, including pre-ringing, as this information would be very useful for optimizing sub systems for multi-listener environments.