CI, TEST: make Travis builds pass again + additional tests in `test_timeseries`

[fkuehlein]

This branch resolves Issue #191 by resolving all remaining flake8 and pylint flags:

• bd798c034fb9884b9b43f3e6994cce17f2ea9f81, b1aeab2d1854e3f96f2f851763f2348af33792af and ddca65f2cfc87e07b4735d013196455fd66c2eed resolve numerous minor flags
• 15d11a89a129cd6d039b6fbcccb70a49360e8cdb resolves too-many-arguments in EventSeriesClimateNetwork.__init__() by handling all optional keyword arguments for parent classes in the **kwargs parameter. kwargs is then split into dictionaries for respective parent class methods in the method body. This way, the behavior of the method remains unchanged from the outside.
• 41c3bdc475035d30d8f3c9b794427e895c1049a2 reworks exception catching clauses by removing them when unhelpful, or adding a re-raise of the caught exception providing additional info.
• fc535f95d1362fcb96253bf685db371986f93def addresses several cases of inaccurate class inheritance that had formerly led to arguments-renamed flags in overriding methods. These have been resolved through realignment of overriding method arguments with their parent class counterparts by way of:
  • grouping additional arguments into tuples/lists or dicts (depending on the context), and checking for the correct number of (sub-)arguments in the method body where needed
  • making all additional arguments formally optional, but checking whether they were given in the method body.

In total, this should hopefully make the Travis builds pass again!

In addition to that, the branch contains additional tests against Issues #145 and #128 in 6b96ced40ef21e2add6393411eb97d5d05194fea, which can hence be closed as well.

Please let me know if you find any inconsistencies or shortcomings in the above described work and will add to it.

Best, Fritz

[fkuehlein]

Update: build errored :(

... but due to an issue with dependency installation, namely

$ travis_retry conda install -c conda-forge networkx matplotlib cartopy sphinx

failing or timing out on all 4 builds because environment cannot be resolved. huh ...

Probably something wrong in the .travis.yaml? Could not resolve it yet, all install commands listed there run flawlessly on my conda.

[ntfrgl]

LGTM! Thank you for this thorough cleanup!

I resolved the problem with the dependency solver by replacing it: https://www.anaconda.com/blog/conda-is-fast-now

[fkuehlein]

Yay, thank you so much for your work! The CI speedup is remarkable and should be of great use in saving runtime credit with Travis.

I will take care of the remaining pylint warning concerning cartopy.