Also, to the OP, is this a 90+ 7/9 with yazaki cluster?
If yes, and if the idle slowly dips ever more, check the grounds by the headlight and by the ECU.
Also, a loose connection/screw for the ground for the dimmer rheostat ground-dump can cause all kinds of strange issues with the clusters on those 93-94 cars.
That said, I've had the later ECUs lower the idling speed and hunt with the lambda feedback on the LH2.4 940 turbos cars more than others.
IDK if its to do with the programming, charging voltage, and fact that the engine almost stalls when the electric fan turns on or what from the factory with the car having no other discernible faults, even if the gauges all read correctly etc/car in good health, no vac leaks, no other known issues.
Cleaning/adjusting the throttle body seems to help a little.
My very lazy theory (hypothesis?/very non controlled environment laboratory 'consistently' repeatable experience?) is just that the AMM is a long way from the throttle body on turbo cars and there's no other major input for load with LH2.4 (so load change detection fairly delayed), the cooling fan speed 1 draws a lot of current to start the fan, & the E-fan cars run pretty ridiculously hot before the ECU finally kicks the fan on & the ECU program lowers the idling speed ever so slightly more the more the temp rises without the program cutting off after the car reaches thermostat temp, so they idle dangerously/raggedly slowly with the A/C on &/or left sitting on a hot day. (fan cuts on at what? 207?F, off at 215?F?).
In the case of -91 (& rest-of-world non-A/C cars that retained the mech fan up to 1998) models, the primary fan is mechanical and slowly progressively locks up; no big draw & those cars in good repair run a little cooler/closer to the ~188? T-stat temp, so they idle a little faster (50-100RPM) from what I've observed with the digital tach cheapie-timing-light with the same ECU either way (the ECU is functioning as programmed).
They also didn't revise ECU programming so heavily for emissions in those years...by the last of them they really had them heaped with EGR etc to squeak by CA for turbos cars and turbos don't have air injection/complex/costly smog pumps on the 8V turbo volvo, so tall order to get them to pass low load/idling running at the razor edge with the 8V head/T-cam even with low compression compromising between lean burn & NOX with the injection system/resolution/programming time they had/could afford for models they were about to kill off in the USA market.
In the case of the 850s/WWDs, they have either a hot film AMM and variable resistor TPS or also have a MAP signal that reacts much more rapidly than the hot wire AMM as the sole significant input to calculate load. The brain is also probably enough 'smarter' to idle up slightly when engaging the fan &/or A/C clutch or has more programming revisions to better keep things running smooth & steady, not to mention finer spraying injectors, sequential injection and just higher resolution/more data points for every input from crank/cam signal to everything else.
They were also designed from the start/clean slate with E-fans across the board, electrical system to go with, A/C & modern emissions for all markets in mind by the time we get to the WWDs for the most part with only some variation (optional air injection pump 1997-1998 on all models, even turbo). & there was more programming experience and more advanced electronics from the get-go/those costs were factored in for the new model from inception rather than a revision/band-aid for an old model that was about to be killed off for the USA market.
This is just my hypothesis for the 92-95 e-fan 7/9s, but it's above my pay grade/mental & physical /monetary resources to prove it. Can say older or newer don't seem to have this run-ability oddity.
Your brain burns hotter than mine, you could test this little, possibly hokum pseudo-science pet theory, but if left to idle, I can say a lot of super low mile in good repair 92-95 factory turbo cars run more or less this way, particularly 94-95 with the last revision ECUs that had to pass emissions most stringently of the RWD 8V turbo cars.
It drove me a little crazy too, as even with the T-cam/healthy compression/drawing plenty of vacuum and all, the somewhat slow, ragged edge of lean, hunting on the lambda switching idle (still basically smooth/doesn't have a miss) and near stall as the A/C & fan engaged was a regular irritation, particularly in hot weather if the car was left to heat up &/or with the A/C on.
Since, tuned up/no detectible faults, the newer WWDs and older -'91 mechanical fan cars wouldn't behave that way/seemed to run no fuss/seamlessly for smooth steady idle/no stalls.
I could never say there was anything "wrong" with all those cars with certainty, just bothered me that they all seemed to run more or less that way, and more-so on warm days/heat soaked with the A/C on. They wouldn't outright stall when the fan turned on with the T-cam though.
Why should the last-revision of the RWD cars that are theoretically 'better' run / behave worse in some ways than their older counterparts?
So many assumptions, clear as mud, right?
Wish you luck, can't see/observe as you do/are from your end, but if you can't find anything else, that (repeated?) experience may have *some* relevance, not sure there's more to add to get the factory gauges/senders working as they should within ohm/output voltage range.
Or maybe I just mis-diagnosed an easy/obvious fault in any/all of those cars or didn't follow some update or TSB...didn't work at a dealer and wasn't properly trained to repair those from new-present en masse, so take it with a grain of salt for sure.
My hack fixes/shotgun attempts:
-Install passenger side radiator temp switch plug in lowest available temp/wiring setup from regina e-fan car (88-92?C?) or saab T with lower temp switch (uglier, but nice that the fan comes on if there's a little coolant loss being near the bottom of the system) in bottom radiator hose to turn on fan. (effective/ran better/idled similar speeds to the -91 mech fan cars in good repair).
-Increase base idling speed somewhat at throttle body such as that's possible on 2.4/install thumb-screw 2.2 throttle body without allowing fast idling coming to rest off the freeway.
-Install mech cooling fan/pusher fan combo from earlier car.
*may not have passed emissions quite as clean, not sure if I have repeated emissions dyno numbers to back it up...didn't keep data that comprehensive.
Maybe this is all irrelevant now and someone can just hack the ECU with a few key strokes/EPROM socket to turn on the e-fan at closer to T-stat temp ('87C or whatever) now or hack it to not lower the idling speed as much as it does once it reaches operating/thermostat temp if my theory is correct?
I *believe* that the EZK brain controls idling speed/idle motor as well as EGR (if equipped) on the LH2.4 & newer 8V RWD cars, not so much the fuel ECU. Fuel ECU no doubt takes into account inputs from the EZK/idling in closed loop.
Unlike the 2.2 cars (no EGR offered), where fuel ECU controls idle motor, ignition can be mopar MAP based or EZK with the same 2.2 ECU/program, no adaptive memory or OBD function.
Low resolution for ignition, but no adaptive memory and less emission control layers is kind of easier to deal with with 2.2 in some ways.
2.2 cars just have a higher base idling speed with a simple +12V input for A/C idle-up/on to avoid a stall with the pusher fan/A/C compressor engaging/disengaging.
<---Other than lack of OBD & some resolution, prefer 2.2 for long-term reliability/tune-ability/low fuss for that era of old hot-wire LH systems.