TST: NEP 029 compliance

[jklenzing]

Description

Adds a test configuration to Travis CI to run NEP 029 recommended minimum support case with Python 3.6 and numpy 1.15. Also removes manual pip installs of pandas, xarray, matplotlib required for python 3.5 testing.

Type of change

• Test environment update (non-breaking change which adds functionality)

How Has This Been Tested?

Updates to Travis CI env, so tested there.

Checklist:

• Make sure you are merging into the develop (not master) branch
• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• Any dependent changes have been merged and published in downstream modules
• Add a note to CHANGELOG.md, summarizing the changes

[jklenzing]

Latest version of numpy is 1.17.2. We are leaving a hole open for 1.16, as well as python 3.7/3.8 with older versions of numpy.

Also, not currently testing older versions of numpy on appveyor.

Should we add more test cases?

[codecov-io]

Codecov Report

Merging #106 into develop will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff            @@
##           develop     #106   +/-   ##
========================================
  Coverage    97.30%   97.30%           
========================================
  Files            4        4           
  Lines          334      334           
========================================
  Hits           325      325           
  Misses           9        9           

---

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 1f43261...ce6e4e1. Read the comment docs.

[JonathonMSmith]

Should we add more test cases?

If we add more test cases, would we test all the permutations?
python 3.6-3.8 with numpy versions 1.15 - 1.18.

Each test takes about 2 minutes (rounding up a lot).
so this would take testing up to about 15-20 minutes.
I think that the testing of the minimum and latest versions is essential.
Other than that I'm not sure if half measures are worth it.
But it could get out of hand down the line.

[jklenzing]

Yeah, on appveyor it will be worse. I think this might be good enough.