Per the OP's OP...
Man, I gotta grin in a weird way as I was beating that drum waaaayyy back, way back before even the old Strong Signals site was in its heyday.
Originally, for me, it was broad-as-a-barn door IF selectivity rapidly followed by lower-than-barely-acceptable (barely acceptable out in the middle of nowhere 100 or more miles from the nearest city) dynamic range and barely-acceptable-if-at-all-present front end filtering. Then were added "Noise Figure?! We donneednostinkin' Noise Figure!!" And freq. stability of LO, poor shielding of stages....and on and on.
For the front end up to the first down conversion:
Uniden determined the least they could get away with and still make it work in many cases and went with that.
GRE determined the least they could get away with and still make it work in many cases and then took it down a few more notches and THEN went with THAT. Well it works out in deep space between galaxies...
For the first IF through to the baseband demodulation:
Initially it was kinda the same but then, from the PSR500 onward GRE got a little more interesting and actually included an almost decent narrow IF filter selection. It didn't really work in P25 mode (you could select it but it actually had no effect) nor in trunking but really helped in analog conventional signals...with some unfortunate side effects (squelch circuit issues, for example); also it wasn't available in search or Tune modes - could only be selected when directly programming a frequency into a memory. But it actually did make real separation of the narrower "narrowband" FM signals possible and practical unlike the Uniden's except for the x36 series. Still, then they went and completely neglected the audio processing of same opting instead to rely on a simple "audio boost" function for the user to manually select. Uniden, on the other hand, again with the exception of the x36 series, stuck with a fairly wide IF that is near useless when presented with a busy swath of closely spaced NFM signals unless you consider it "convenient" to just punch in one central frequency and hear everything +/- 25kHz...guess you could sorta-kinda consider that a "feature"...kinda sorta maybe. BUT...! Hey, unlike Uniden they DID get the audio processing right! Face palm and head shaking...go figure...
I think there was a time when "base units" differed significantly from their mobile and portable counterparts but that has long past. Now, in the interest of reducing costs and simplifying design efforts the usual approach is to make one primary RF/IF/Baseband design optimized for portable use and then use that in a mobile/base as well as a portable form. Primary difference between forms is audio amplifier design and power supply. Yes, you could use better filtering, shielding, and more power hungry RF amps with better dynamic range in a fixed larger size case as opposed to a smaller battery operated portable unit but then that would mean added design time, more and different parts selection, factory build time, and overall higher cost that just is no longer acceptable in a low margin niche consumer market.
LSM...well I've also beat that to death myself here as many know. I recall scratching my head at the use of a discriminator for P25 demodulation when I first found out about that design in consumer scanners. But, to be fair, I was in error as P25 C4FM (non-simulcast) is a constant envelope modulation which can be handled fine using a discriminator approach design. I came from an engineering background in DAMPS pi/4 CQPSK and was incorrectly looking at P25 Phase 1 from that point of view. But, when simulcast issues began rearing their ugly heads, it started to look relevant again.
Simply put - a true I/Q demodulator can decode anything, AM, FM, SSB, C4FM, CQPSK, QUAM, whatever, with the proper adjustments, etc. A discriminator is limited to constant envelope angle modulated signals only (well, technically above a certain low signal level spec.).
And that is with hardware designs.
Today, it is obvious to me that the software back end is the way to go going forward. OP25 proves this. Build your I/Q demodulation in software and feed it with your final hardware downconverted IF and damn near work bloody magic! Still requires a decent hardware RF front end and IF downconversion with all of the filtering, LO stability, noise figure, dynamic range, and shielding considerations for best performance but the flexibility in the demodulation is near bliss compared to what I had to work with twenty years ago.
I think the primary reason manufacturers of consumer scanners don't (or didn't) use hardware I/Q demodulation circuits is that there were/are no low cost large scale antenna-to-audio-amp RF IC's suited for general purpose wideband receiver applications. To keep costs and design time down they relied on such large scale system-on-a-chip based designs. Outside of very specialized IC's which were limited to certain fixed band uses like cellular phones and WiFi, there were only the much higher cost professional IC's which were/are probably limited in distribution to designers of professional gear.
Now with the magic of SDR approaches, however, a different design potential presents itself. Until we have, however, super wideband A/D converters with super high dynamic range it looks to me like the best approach is to use a combination of tried-and-true RF hardware for the antenna to first or maybe final IF and SDR with a properly implemented I/Q demodulator.
-Mike
