Ya know, looking at your closed loop scan, something struck me...
There is some debate on the tuner forums as to how the VE table in the PCM is used in a MAF-enabled engine.
There is one camp that says the VE table and MAF data are somehow "blended together" by the PCM at rpm below 4000, after which MAF takes over entirely.
The other camp says that VE table is simply a "backup system" in case of MAF sensor failure.
I gravitate toward the second camp because I have done targeted experiments in my OBDII '98 L36 where I screwed around with the VE table on purpose to try to cause a "reaction" in a MAF-tuned tune...and I could get none.
The first camp I always thought of as "urban legend" believers or perhaps "holdovers" from the LT1-edit days; however, perhaps indeed your '95, OBDI (-ish) PCM does
do some kind of VE table / MAF blending once in closed loop. I say this because I note that in your closed loop scan, the MAF vs. rpm "discontinuity" I see occurs at just about 4000 rpm.
What this has me thinking is that perhaps your PCM'* VE table just cannot deal with your cam, and simply has to be re-programmed. Why this would cause
screwy MAF readings I don't know...and unfortunately, this is where my usefulness ends, 'cause I have no experience programming an OBDI system.