The car is 1997 Civic DX (D16Y7 non-vtec engine), everything stock.

Problem: Runing rich at idle (STFT=12-13%, LTFT=3%) and low RPM

- Car runs consistently rich on idle and low RPM with stable STFT at 12-13% and stable LTFT at 3%.
- High RPM brings fuel trim much lower (2-3%) with STFT swinging from positive to negative.
- This engine has MAP sensor so no vacuum leak on intake side or idle RPM will be higher.
- I was considering vacuum leak on exhaust side (there is welded crack on exhaust manifold before the upstreal O2 sensor), but upstream O2 sensor says this is not the case as it indicates rich.
- The exhaust pipe CO emissinos are high (from last smog check at 1500RPM, CO is still a little under limit), and HC are low - so running rich, but no misfire at all (and no misfire indications from PCM).

Details from OBDII live data (car fully warmed up as ECT shows):
IAT: 96 degrees F (good)
MAP: 3.5 PSI (within specs)
BARO: 14.4 PSI
TPS: -0.6 degrees (???)
HO2S S1: 0.62V (swinging lightly, mostly above 0.450V -> indicating rich mixture)
HO2S S2: 1.23V (???)
RPM: 675
ECT: 180 degrees F
STFT:12
LTFT:3
F INJECTOR: 2.9ms

So what I have is:
1) upstream O2 sensor is not the reason for the high STFT (already indicating rich mixture), something else is overpowering it. At high RPM upstream O2 swings between 0.2 and 0.8V a couple times per second, which indicates good O2 sensor
2) downstream O2 sensor -> does not swing, stable slow changing voltage. This started at 3.25V at engine start (actually both upstream/downstream started at 3.25V ???) and slowly went to 1.23V at fully warmed up engine. I thought that O2 voltage is bounded at 0-1.25V and 0-1V for warmed up engine. Is this a sign of bad sensor? Does downstream O2 affect PCM on this engine (most do not use downstream O2, but some do use them).
3) TPS -> -0.6 degrees -> is this a bad sensor? (service manual says 10 degrees at idle)
I thought that bad TPS reading can send PCM into "safe mode" where it would maintain STFT/LTFT constant (12/3) to avoid stalling. Unfortunately does not seem to be the case as TPS voltage reading at PCM is within specs (0.5V at idle, 4.7-8V at WOT and increasing in the middle).

Questions:
a) why PCM demands rich mixture when upstream HO2 indicates rich already?
b) why STFT stays at 12% consistently and LTFT does not adjust at all?
c) TPS: -0.6 degrees -> is this wrong? could it be the reason?
d) downstream O2 sensor -> what does this indication mean? Bad sensor?

Thanks!