[WIP] A couple more commits on top of Matt and Christians msmbuilder2.8 work

[rmcgibbo]

I don't think it's possible for me to add onto Matt's PR #264, so i just built off it.

[rmcgibbo]

----------------------------------------------------------------------
Ran 109 tests in 17.871s

FAILED (SKIP=5, failures=7)

[rmcgibbo]

"Showing 126 changed files with 733 additions and 69,602 deletions."

[kyleabeauchamp]

OK I read through everything to get a feel for what's going on here, looking good...

[rmcgibbo]

Thanks for looking through all that stuff.

[rmcgibbo]

The sphinx documentation actually looks pretty good now.

[rmcgibbo]

I think this is ready to merge, modulo a couple test failures. I am not really sure if the test failures are significant, or if they're just testing aspects of msmbuilder2.6 that were sort of random defaults. Or some subtle numerical differences. For instance, the test_Cluster_kcenters gives `Arrays are not equal, mismatch 0.00199%)

[rmcgibbo]

Also, CreateAtomIndices.py gives an extra atom index for ala2.pdb due to a bug in 2.6. In .lh5 format, the atomName for the terminal NME carbon is CH3, which doesn't get included in the atom indices. MDTraj's pdb reader normalizes the name to C, and so it gets included.

[rmcgibbo]

Here's a link to the new documentation: http://rmcgibbo-msmbuilder.readthedocs.org/en/latest/

[kyleabeauchamp]

Hmm. So I would say that the current docs suggest that everyone should be using SCRE, which I'm not sure is the case. Any way to put it further "down" the hierarchy?

[kyleabeauchamp]

I could play with this later, as well--it's probably a separate PR that I should do.

[kyleabeauchamp]

OK the docs look really sweet. I'm happy.

[rmcgibbo]

Yeah, I just ran stuff through pandoc and didn't do anything with the content. There's still an issue with getting the autosummary to work on RTD.

[rmcgibbo]

I'm merging this into msmbuilder/msmb2.8. It's still not the master branch here.