Add Bayesian Blocks binning functions for binned and unbinned data

[AdamGoldstein-USRA]

A highly requested feature is the Bayesian Blocks binning algorithm implemented in the GDT.

This could be implemented using Astropy's bayesian_blocks method. Both binned and unbinned data would use fitness='event' and would accept the gamma and ncp_prior keywords.

For unbinned data, the GDT function would be little more than a wrapper around bayesian_blocks, since the input is an array of event times and the ouput is an array of bin edges.

For binned data, the GDT function is slightly more complex because the number of counts in each new bin must be calculated as well as the exposure of each bin.