PyMBAR (4) will create an endless loop in `get_forward_and_reverse_analysis` if there are too few samples

[IAlibay]

Describe the bug

If you run a short simulation, there is a risk that you might end up with a situation where 10% of the simulation has too few samples for MBAR analysis. This leads to a convergence failure which loops endlessly (or at least doesn't resolve in the span of several hours).

To Reproduce

Run a ~ 500 ps AHFE simulation.

Expected behavior

Possible ways to get around this case:

• Try to automatically rebalance how the forward/reverse analysis is chunked so that we have sufficient amounts of samples in each chunk.
• If we have too few samples in each chunk (unsure what a good number would be, less than 50?) and return None.

[mikemhenry]

Can you also report this upstream? We can address it separately/have a work around but it would be good to loop upstream into this