MAINT: replace progressbar with tqdm

[fkuehlein]

Deleted the utils/progressbar and everything that relates to it. Added tqdm dependency and replaced all ProgressBar instances with tqdm progress bars.

@ntfrgl, I'm not entirely sure I added the tqdm dependency in all the right places in setup.cfg and .travis.yml, could you please double check?

[codecov]

Codecov Report

Merging #202 (585c503) into master (ea721c1) will decrease coverage by 0.19%.
The diff coverage is 22.72%.

@@            Coverage Diff             @@
##           master     #202      +/-   ##
==========================================
- Coverage   52.04%   51.85%   -0.19%     
==========================================
  Files          49       45       -4     
  Lines        7473     7064     -409     
==========================================
- Hits         3889     3663     -226     
+ Misses       3584     3401     -183     

Files	Coverage Δ
src/pyunicorn/climate/havlin.py	61.36% <100.00%> (-2.85%)	⬇️
src/pyunicorn/climate/mutual_info.py	71.25% <ø> (+19.49%)	⬆️
src/pyunicorn/timeseries/surrogates.py	51.89% <50.00%> (+1.63%)	⬆️
src/pyunicorn/core/network.py	54.64% <11.11%> (+<0.01%)	⬆️

[fkuehlein]

CodeCov is complaining that overall test coverage is actually slightly reduced. Apparently, the few additional lines in untested methods outweigh the countless untested lines I removed. Not sure I understand the logic, could add tests of course, but I'd say that's a different issue.

[ntfrgl]

I would inline this into the trange() call.

[ntfrgl]

This obviously applies equally to the other changes.

[ntfrgl]

This coverage metric ended up slightly lower, because the coverage within the old utils/progressbar, which was entered via the removed lines in the tested methods, was somewhat higher than the current overall average over pyunicorn. Irrespectively of the particular metric, this PR is an improvement to the project code base.

For the record, the following are the untested methods in which the progress bar was replaced, and which appear not to be used anywhere else in the project. The last one in particular has been replaced by its Cythonized version, so I suggest deleting it.

• Network.GrowPreferentially‎()
• Network.GrowWeights‎()
• Surrogates‎.test_threshold_significance‎()
• MutualInfoClimateNetwork._calculate_mutual_information‎()

I would also suggest to start writing the change log for the new release. This PR would be represented by a one-liner there.

Otherwise, this looks good to me.

[fkuehlein]

Thanks for the hints!

I improved the silencing of the new progressbars accordingly and removed MutualInfoClimateNetwork._calculate_mutual_information‎(). Made CodeCov slightly happier, too!

Wouldn't want to decide about the other apparently stale methods myself. Maybe @jdonges can have a quick look? Shouldn't keep us from merging this though.

Oh and I started taking notes for the changelog now, good point.

[ntfrgl]

Thank you.

In contrast to MutualInfoClimateNetwork._calculate_mutual_information‎(), the other untested methods are public, so I don't see a reason for removing them at this point.