Add CI based on tox (github workflows and templates based on oemof.solph)

[joroeder]

This PR suggests tox as test environment, and adds the following github workflows (based on oemof.solph):

• lint (flake8, I removed black - however, we could also add this later, but we were using flake8 in DHNx before)
• packaging
• tox checks (maybe we need to add a tox.ini first ;-)
• pytest, test coverage (also with tox)

Is there anything against using the github actions?

Any changes required? Any comments? - I am not the expert of CI ;-)

• [x] Make pytest running
• [x] Enable isort again ?
• [x] flake8: include examples again ?

• [x] Add hint that tox is used in the documentation

  Locally, you can test with:

  tox -e "clean, check, docs, py38"

[joroeder]

@jnnr It seems that within the test environment, there are problems with the files tests/_files/tree_network_sim_expected_results/calc_pressure_loss.py and tests/_files/tree_network_sim_expected_results/calc_temp_drop.py, that refer to data in the examples folder, that cannot be found (see also test_log_tox). I wonder why there are .py files in `tests/_files/*`` at all? Are these scipts needed? I mean for generating the data files, that are checked within the tests, the scripts for generating the data do not need to be there, right?

Concerning the examples, there are some examples, that generate output files in the respository, that are also not excluded by .gitignore. I commented already two cases (Case1, Case2), as I had problems with the CI, that should be setup with this PR.

Could you have a look at that?

Feel free to directly commit into this PR

[joroeder]

@jnnr It seems that within the test environment, there are problems with the files tests/_files/tree_network_sim_expected_results/calc_pressure_loss.py and tests/_files/tree_network_sim_expected_results/calc_temp_drop.py, that refer to data in the examples folder, that cannot be found (see also test_log_tox). I wonder why there are .py files in `tests/_files/*`` at all? Are these scipts needed? I mean for generating the data files, that are checked within the tests, the scripts for generating the data do not need to be there, right?

There were old parameter names (diameter_mm, length_m), I fixed them, and the scripts are running. I am still wondering why there are the files? Maybe, we should at least remove the link to data in the examples (e.g. if the examples are moved in future, the tests raise errors.) - separate issue/PR?

Concerning the examples, there are some examples, that generate output files in the respository, that are also not excluded by .gitignore. I commented already two cases (Case1, Case2), as I had problems with the CI, that should be setup with this PR.

We could do this also in a separate PR, as there are no conflicts with the CI at the moment.

[joroeder]

From my side, this PR is done. (besides resolving the remaining comments and questions)

[joroeder]

Hey @jnnr

I completed the open things we discussed:

• Add examples output to gitignore
• Make API documentation be built on ReadtheDocs
• clean extras_require (I now kept one options "tests", that includes all extras_require but not cartopy, that did not work in the github workflow)

As everything looks fine now, I will merge this PR.