As you mentioned, the simple FM demod is how the PRO96 appears to work, according to the schematic.
The Motorola units use an abacus chip and actually digitize the baseband for analysis.
I'm not familiar with the "abacus chip" you mentioned but, otherwise, that sounds like the method I am familiar with from working on PI/4 DQPSK modulated TDMA cellular units back in the day. In fact, I had never heard of a "data slicer" at all until I got heavily into these hobby forums. I was always under the impression that I/Q modulators/demodulators were the primary method of dealing with complex digital modulation signals.
The way I remember it, on the receive end, I think I recall that we took the downconverted IF ran it through an I/Q demod IC and then A/D'ed the I and Q signals. They were then processed by DSP engines for equalization, error correction, and demod (voice data extraction, etc.); however, they might have applied the equalization to the digitized final IF before I and Q demod (which would seem to require an extra A/D and, possibly, D/A step...?). It's been quite awhile and I mostly worked on the transmit side at the time.
I would imagine that the operation of the 500 in terms of how it demodulates digital signals is probably unchanged from that of the 96 (at least in the fundamental sense). That's actually a shame but I guess it's cheaper than using the I/Q demod method. The thing is, it seems to me that with cellular phones using it pretty ubiquitously (at least they were when I worked on them) one would think it to be fairly cheap to implement in a scanner?!?
Personally, I would like to see them move away from FM discriminator efforts and toward true I/Q demod if they haven't already done so, especially given the move toward increasingly complex digital signals by the industry. Using true I/Q raw output rather than a tapped discriminator would also really benefit those working on open source multi-mode digital voice demodulating/decoding software;-)!
-Mike