The behaviour of trying USER0 or the A-drive automagically when a COM file is not found on the current drive seems odd to me after having used CP/M on different platforms. But it sure seems like a useful feature. Regarding the question if anyone has done something to improve the CCP... Historically, yes, and there's a reason Z-system exists :) However, I don't think there's many "alternate BDOS" versions made for systems based on Searle's BIOS (Searle's system, RC2014, LiNC80).
I started looking at this before seeing you reply Steve, and came to some of the same conclusions as the posting you linked to. I think the location to "sneak in" the change would be in the "UNK*" area of the CCP code, between lines 1080-1180 (approx) of BDOS_22.asm, but it may be tricky. Unless the entire code is shifted to move CBASE down from 0xD000, there's precisely 13 bytes available for this change between the CCP and the FDOS. The "padding" is located at line 1240 in my copy, by the comment "Room for expansion?"
If anyone is up for the challenge, I will happily promote the resulting version.