TST: Make HDF5 fspath write test robust

[TomAugspurger]

The test_write_fspath_all test would fail on the HDF5 example
occasionally (about 1/100 in my experience). Apparently you don't get an
identical HDF5 every single time. This refactors that test out to its own where
we write and read both versions, and compare equality that way.

See #14026 (comment) and https://circleci.com/gh/pandas-dev/pandas/2507?utm_campaign=vcs-integration-link&utm_medium=referral&utm_source=github-build-link for an example

[codecov]

Codecov Report

Merging #16575 into master will decrease coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #16575      +/-   ##
==========================================
- Coverage   90.78%   90.75%   -0.03%     
==========================================
  Files         161      161              
  Lines       51089    51089              
==========================================
- Hits        46380    46368      -12     
- Misses       4709     4721      +12

Flag	Coverage Δ
#multiple	88.59% <ø> (-0.03%)	⬇️
#single	40.16% <ø> (ø)	⬆️

Impacted Files	Coverage Δ
pandas/plotting/_converter.py	63.23% <0%> (-1.82%)	⬇️

---

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 fb47ee5...7b1e9c5. Read the comment docs.

[codecov]

Codecov Report

Merging #16575 into master will decrease coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #16575      +/-   ##
==========================================
- Coverage   90.78%   90.75%   -0.03%     
==========================================
  Files         161      161              
  Lines       51089    51089              
==========================================
- Hits        46380    46368      -12     
- Misses       4709     4721      +12

Flag	Coverage Δ
#multiple	88.59% <ø> (-0.03%)	⬇️
#single	40.16% <ø> (ø)	⬆️

Impacted Files	Coverage Δ
pandas/plotting/_converter.py	63.23% <0%> (-1.82%)	⬇️

---

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 fb47ee5...7b1e9c5. Read the comment docs.

[codecov]

Codecov Report

Merging #16575 into master will decrease coverage by 0.02%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master   #16575      +/-   ##
==========================================
- Coverage   90.78%   90.75%   -0.03%     
==========================================
  Files         161      161              
  Lines       51089    51089              
==========================================
- Hits        46380    46368      -12     
- Misses       4709     4721      +12

Flag	Coverage Δ
#multiple	88.59% <ø> (-0.03%)	⬇️
#single	40.16% <ø> (ø)	⬆️

Impacted Files	Coverage Δ
pandas/plotting/_converter.py	63.23% <0%> (-1.82%)	⬇️

---

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 fb47ee5...7b1e9c5. Read the comment docs.

[jreback]

thanks!

[jreback]

you might want to backport this? up 2 u

[TomAugspurger]

I think the fspath stuff was 0.21.0 (since it had that minor API change on HDFStore's getattr)

[jreback]

ahh right I remember your comment. thanks.