Enhancements

[j1c]

This is a list of things that can be changed to substantially improve performance of m2g. Some of it is addressed in #398

• [x] Im 1000% percent sure below line is tabbed incorrectly. It should be outside the for loop. https://github.com/neurodata/m2g/blob/be29587322ab1fafb96f6afb726efbdb39b64b66/m2g/graph.py#L216

with ~1 million streamlines, it takes ~35 minutes vs 2 minutes with the line outside the for loop.

There is a parallel version in my PR.

• [ ] I think the streamline generation can be parallelized by sending partial seed list to bunch of Track instances.