Voltage of lithium ion also varies depending on load, and is very flat for most of the discharge/charge cycle. The difference is that the endpoints are well known.
I cannot verify one way or the other, personally. But my previous statement has completely different implications re: li ion vs NiMh. And I'm talking strictly float voltage. Never mentioned load.
Two li ion cells that both measure xV float voltage are in a similar SoC. Even though the curve may be flat in this area, you could tell they are both within a certain, contiguous range of SoC. And given enough decimal places and a high enough test probe impedance, you could get a pretty good rough estimate.
Apparently, two NiMh batteries at xV may be in two completely different, isolated points where that SoC curve goes up/down through xV. So at xV volts, SoC might be (for sake of argument) between 20%-30%... OR it might be between 80%-90%. And you wouldn't be able to tell by the voltage. Again, I can't personally attest, but what I understand is that there is at least one "hump" in the voltage-over-SoC graph for NiMh, or maybe it's not even completely consistent between individual cells and/or actually even changes through charge cycles.
Imagine what occurs in parallel charging. In lead acid or li ion, the batteries will end up charged at the same time. The one that may start to lag behind will suck relatively more current, in a self-balancing manner, and they will finish at the same time. In NiMh, once the cells reach one of those negative slopes on the curve, one of them will break through first, and the one with the higher SoC will start to suck more of the current in a positive feedback loop, possibly even sucking some current from the lower SoC cell during this time.
Been many years since I had any interest in NiMh, but this was my understanding of why NiMh do not like to be charged in parallel. Yeah, I can imagine the thermal feedback might be necessary even if the curve were simply particularly flat at that point, I suppose.