Addresses failure of test_load() in PSA (#2049)

[sseyler]

Fixes #2049 and also boosts PSA code coverage (#1006).

Changes made in this Pull Request:

• test_load() now passes after calling save_paths() with default arguments
• PSAnalysis now has default base filename and prefix for consistency with default saving behavior of PSAnalysis.generate_paths()
• Better consistency of Path.fit_to_reference() with MDAnalysis.analysis.AlignTraj
• Better consistency of PSAnalysis.generate_paths() with MDAnalysis.analysis.AlignTraj
• Updated PSA docs for consistency with above changes

PR Checklist

• Tests?
• Docs?
• CHANGELOG updated?
• Issue raised/referenced?

[codecov]

Codecov Report

Merging #2239 into develop will increase coverage by 0.07%.
The diff coverage is 62.5%.

@@             Coverage Diff             @@
##           develop    #2239      +/-   ##
===========================================
+ Coverage     89.6%   89.67%   +0.07%     
===========================================
  Files          159      159              
  Lines        19728    19728              
  Branches      2780     2780              
===========================================
+ Hits         17677    17692      +15     
+ Misses        1457     1440      -17     
- Partials       594      596       +2

Impacted Files	Coverage Δ
package/MDAnalysis/analysis/psa.py	83.49% <62.5%> (+2.87%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f878cc1...b49a6b0. Read the comment docs.

[orbeckst]

Thanks @sseyler , I'll wait for CI to finish. Feel free to ping me.

[orbeckst]

Weird travis failure on macOS 960.10 (Python 2.7) and 960.9 (Python 3.6)

$ eval $MAIN_CMD $SETUP_CMD
/Users/travis/.travis/functions: line 104: pytest: command not found
The command "eval $MAIN_CMD $SETUP_CMD" exited with 127.

even though the installation logs indicate that pytest package was installed:

   pytest-4.4.0               |           py36_1         344 KB  conda-forge

@MDAnalysis/coredevs any idea what's wrong here?

[orbeckst]

This passed CI on Linux and Windows but macOS has a weird failure not related to this PR so I am approving it for right now. Thanks @sseyler !

[tylerjereddy]

I pushed this feature branch to my fork and opened a PR to the same master branch on my fork & so far the Mac job there is doing just fine with pytest running the suite: tylerjereddy#2

Maybe a restart is warranted, and if it starts showing up more regularly we can open an issue.

[tylerjereddy]

Commit history isn't the cleanest, but CI is all green now if that's what you were waiting for.

[orbeckst]

Many thanks @tylerjereddy . I'll quickly compact the history a bit.

[tylerjereddy]

We're getting the MacOS failures again--have those happened in any PR other than this one?

[orbeckst]

yes:

• recent merge into develop 965.9 (merge of PR Fixes transformation tests not running. #2243)
• this PR (I just restarted the macOS jobs)
• others?

[orbeckst]

I have to resolve the merge conflict in this PR so CI will run again anyway...

[orbeckst]

Now AppVeyor doesn't want to run... will rebase manually.

[orbeckst]

macOS still failing