address all-zero time-dependent value removal bug (closes #722)

[fneum]

Still have to check how much it affects file size.

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: -0.04% ⚠️

Comparison is base (fd30d88) 77.92% compared to head (432ad72) 77.88%.
Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #734      +/-   ##
==========================================
- Coverage   77.92%   77.88%   -0.04%     
==========================================
  Files          26       26              
  Lines        6795     6793       -2     
  Branches     1479     1478       -1     
==========================================
- Hits         5295     5291       -4     
- Misses       1182     1185       +3     
+ Partials      318      317       -1     

Files Changed	Coverage Δ
pypsa/io.py	89.71% <100.00%> (-0.39%)	⬇️

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[coroa]

Fixes #722

[coroa]

I agree this makes sense for the netcdf representation, where there is compression (which is now enabled by default), but there are other exporters like to csv where this has a big impact (not only in storage space, but also in file space, which incidentally is also where this branch was originally implemented and included into the exporter/importer infrastructure).

Yes the interaction with optional attributes like p{n} was not considered at the time (since that came later), but you should still test out the drawbacks of including everything clearly.

(Moved here from issue)