Adaptive mask is generated using whole acquisition matrix, instead of brain mask

[tsalo]

Summary

The adaptive mask identifies its exemplar voxel(s) from the 33rd percentile of the available data. Currently, the available data used include the whole acquisition matrix, but it should be limited just to the brain mask.

Additional Detail

Stems from #1057 (comment).

Next Steps

1. Test out current and proposed adaptive masking approaches on a range of datasets to see what effect it has.

[tsalo]

Here are results of the two approaches on some five-echo data I had on hand.

Original

This is after applying the brain mask after calculating the adaptive mask. I can plot the adaptive mask without applying the brain mask if folks want. Just let me know.

Limited to mask