ENH: support pathlib in ReaderBase

[tylerjereddy]

• the creation of a Universe object involving GROMACS formats should now accept pathlib
  objects as requested in Using pathlib Path object #2497; resolving that issue entirely is a separate matter, this branch focuses only on the GROMACS format and only if you proceed via the public Universe creation API

• probably could make it work in Cython directly, though I'm not convinced we need to do that for this part of the fix at least, and we probably should still support strings as well of course

[IAlibay]

Shouldn't this be a ReaderBase thing instead to store self.filename properly?

[codecov]

Codecov Report

Base: 94.36% // Head: 94.36% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (af487bc) compared to base (497f45e).
Patch coverage: 100.00% of modified lines in pull request are covered.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #3935   +/-   ##
========================================
  Coverage    94.36%   94.36%           
========================================
  Files          194      194           
  Lines        25109    25115    +6     
  Branches      3398     3399    +1     
========================================
+ Hits         23694    23700    +6     
  Misses        1366     1366           
  Partials        49       49           

Impacted Files	Coverage Δ
package/MDAnalysis/coordinates/base.py	94.63% <100.00%> (+0.02%)	⬆️
MDAnalysisTests/coordinates/base.py	94.28% <0.00%> (+0.02%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[hmacdope]

Shouldn't this be a ReaderBase thing instead to store self.filename properly?

Seconding this, can this not be dealt with at ReaderBase level to fix it for all readers?

[tylerjereddy]

Ok, I moved the patch to ReaderBase and added a test for one more format (DCD) as a sniff test for generalization.

I think that's probably good enough for now, but note that:

1. This still doesn't cover all readers because SingleFrameReaderBase has its own self.filename -- I'd be happy to delay patching/testing that for now.
2. The testing could probably be centralized/parametrized somewhere, covering a bunch of formats rather than testing the same pass-through for just two formats. Maybe good enough for now though..

[IAlibay]

Assuming this passes tests I'm happy for this PR to go through. @tylerjereddy could you raise an issue about the missing bits? Maybe we should move stuff to ProtoReader in the future but it seems fine for this current issue?

[tylerjereddy]

Eh, the failure is related--this is the problem with inheritance... I'd just as soon scope this back to GROMACS and move on.

I'll see if I can add a shim to fix it though.

[tylerjereddy]

Ok, I pushed in a commit to shim around our NamedStream stuff in the base class, and this seems to allow the full suite to pass for me locally at least.

gh-3937 contains the the follow-up reminders I noted above.

[hmacdope]

LGTM @tylerjereddy

[tylerjereddy]

Ok, I have approvals from 3 core developers and CI is passing, in it goes.