CI: troubleshoot UnsatisfiableError in travis-36-cov on 1.1.x

[simonjayhawkins]

• closes #xxxx
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[simonjayhawkins]

@TomAugspurger the coverage job on travis first timed-out here https://travis-ci.org/github/pandas-dev/pandas/builds/728369353

I can replicate with just nomkl, scikit-learn and numpy[1].

AFAICT the versions of these have not changed since the last good build. Any ideas?

output from local below. running here to see if same.

UnsatisfiableError: The following specifications were found to be incompatible with each other:

Output in format: Requested package -> Available versions

Package setuptools conflicts for:
scikit-learn -> joblib[version='>=0.11'] -> setuptools
python=3.6 -> pip -> setuptools

Package numpy conflicts for:
scikit-learn -> numpy[version='>=1.11.3,<2.0a0|>=1.14.6,<2.0a0|>=1.16.5,<2.0a0']
numpy=1.15
scikit-learn -> scipy -> numpy[version='>=1.15.1,<2.0a0']

Package python conflicts for:
python=3.6
scikit-learn -> python[version='>=2.7,<2.8.0a0|>=3.6,<3.7.0a0|>=3.8,<3.9.0a0|>=3.7,<3.8.0a0|>=3.5,<3.6.0a0']
scikit-learn -> joblib[version='>=0.11'] -> python[version='2.7.*|3.5.*|3.6.*|>=3.6|3.8.*|>=3.5|3.7.*']

[1] Updated: building the environment will also fail if excluding numpy. but we want that pinned.

[simonjayhawkins]

closing, not getting same error as local here.