Reduce Scatter Block issues at large scale

[wckzhang]

Reduce Scatter Block hits lots of issues in regards to running out of memory and such at ranks 128+. Need to figure out a way to handle this...Possible to just ignore the failures and just revamp the parsing code a little. Needs more thought.

[wckzhang]

This is more of a robustness issue, non critical but it's annoying since I have to juggle two datasets, one with reduce_scatter_block, and one with every other collective.