yaml output files

[Duncan Ralph]

I just merged into master the new default yaml output file format. This is a big simplification, since it allows the germline set, annotations, and partitions to all be in the same file. It's hopefully explained well enough here (example here).

a couple notes:

  - it's backwards compatible, in that if you keep specifying .csv output files, you'll keep getting the same output files you've gotten in the past.

  - it's actually reading and writing with the json subset of yaml, since json.dump/.load are much faster than their yaml alternatives. A reasonable argument could be made that I should've just used json, but yaml is more human readable by non-computer people (i.e. if you swap a json.dump for a yaml.dump with nice formatting, a non-command-line-ey person could probably guess the info they want by eye), and it's nice to be consistent with the existing hmm model files, which are already yaml.

  - while it's been pretty thoroughly tested, I was planning to run for a few more weeks before merging, but Branden found an unrelated rare but highly annoying crash that needed fixing. So, if you see something funny, do please open an issue.

thanks