On the contrary Triplex, you've proven that point with your exceptions.
The NYC's 0-8-8-0s were not road engines. Considering that it was a race track road...they would have looked horribly out of place. Unless someone is modeling a hump yard (or a rr museum)...they'd make no sense.
Further, the Niagaras were not big northerns. They had lower tractive effort than NKP berks (but higher HP...the niagara's were more than a decade newer in design)...and they were not particularly heavy. They essentially combined the mohawks with efficiency improvements...namely the 4-wheel trailing truck...and tall drivers. They also were stretching the limit of what could operate on the water level route due to the NYC's tight clearances.
They were, to a degree, experimental power much like the Pennsy's T-1s & Q-2s (and odd balls) which looked out of place on a railroad which prided itstelf on a fleet of plain, low tech 2-10-0s, 2-8-2s, 4-6-2s, etc.
Further, while roads did pick up equipment through mergers, normally they would purge their rosters (as they could afford to do so) to streamline their maintanence costs. And even if the didn't purge it, the usually wouldn't send their new equipment all over their system. The W&LE 2-6-6-2s didn't find themselves travelling over the NKP's mainline after 1949...they stayed on their own trackage. Same thing with the W&LE berks...even though they were essentially copies of the NKP's berks (but with ugly boxpoc drivers).
In fact, most power was division specific. The C&O operated distinctly different power over each division. Even with identical power...the NKP divided up its power...the 765 always worked in northern Indiana...the S's worked the LE&W...and they designated 4-6-4s to the mainline...4-6-2s to the St. Louis line...and 4-6-0s to the Peoria passenger trains.
As for merging...the british rail un-merger resulted in the exact kinds of odd ball rosters that you mentioned...one of the worst-turned best sucesses...wales & west if I recall...hired an american to come over...her dumped the hodge podge collection and bought all new power from GE or GM to standardize...which resulted in substantially lower operating costs.
Just because Penn Central had 80 differend models doesn't mean that they actually used hardly any system wide...and therefore...only a small fraction would be relevent to any modeler unless they were modeling the entire system.
Just because a complete roster was diverse doesn't mean that it would be relative to a model railroad. On the other hand, modeling a division point would make a diversity of power plausible...IF each division had different operating characteristics. And then the freight crews and power would need to be changed...for nice operating characterists
