Fix failing conda tests

[GeorgianaElena]

The tests are failing for some time :( However, the strange thing is that the tests in test_conda.py all pass if ran individually.

At first I thought this might be related to the newer version of pytest (since the tests passed if each of it got its own instance of the prefix fixture). But there wasn't any pytest change to justify the behavior.

After more digging, the issue I noticed is that after the first conda.ensure_conda_packages call, the environment got modified and instead of having symlinks to python3.6, there were instead to python3.7, hence the "ModuleNotFoundError: No module named 'conda'" error.

conda-meta/history contents before the running any test

+defaults::python-3.6.5-hc3d631a_2
+defaults::ca-certificates-2018.03.07-0
+defaults::conda-env-2.6.0-h36134e3_1
+defaults::libgcc-ng-7.2.0-hdf63c60_3
+defaults::libstdcxx-ng-7.2.0-hdf63c60_3
+defaults::libffi-3.2.1-hd88cf55_4
+defaults::ncurses-6.1-hf484d3e_0
+defaults::openssl-1.0.2o-h20670df_0
+defaults::tk-8.6.7-hc745277_3
+defaults::xz-5.2.4-h14c3975_4
+defaults::yaml-0.1.7-had09818_2
+defaults::zlib-1.2.11-ha838bed_2
+defaults::libedit-3.1.20170329-h6b74fdf_2
+defaults::readline-7.0-ha6073c6_4
+defaults::sqlite-3.23.1-he433501_0
+defaults::asn1crypto-0.24.0-py36_0
+defaults::certifi-2018.4.16-py36_0
+defaults::chardet-3.0.4-py36h0f667ec_1
+defaults::idna-2.6-py36h82fb2a8_1
+defaults::pycosat-0.6.3-py36h0a5515d_0
+defaults::pycparser-2.18-py36hf9f622e_1
+defaults::pysocks-1.6.8-py36_0
+defaults::ruamel_yaml-0.15.37-py36h14c3975_2
+defaults::six-1.11.0-py36h372c433_1
+defaults::cffi-1.11.5-py36h9745a5d_0
+defaults::setuptools-39.2.0-py36_0
+defaults::cryptography-2.2.2-py36h14c3975_0
+defaults::wheel-0.31.1-py36_0
+defaults::pip-10.0.1-py36_0
+defaults::pyopenssl-18.0.0-py36_0
+defaults::urllib3-1.22-py36hbe7ace6_0
+defaults::requests-2.18.4-py36he2e5f8d_1
+defaults::conda-4.5.4-py36_0
# update specs: ['conda', 'python=3']

conda-meta/history contents after the running first test (conda.ensure_conda_packages(prefix, ['numpy']))

-defaults::ca-certificates-2018.03.07-0
-defaults::certifi-2018.4.16-py36_0
-defaults::libffi-3.2.1-hd88cf55_4
-defaults::libgcc-ng-7.2.0-hdf63c60_3
-defaults::ncurses-6.1-hf484d3e_0
-defaults::openssl-1.0.2o-h20670df_0
-defaults::pip-10.0.1-py36_0
-defaults::python-3.6.5-hc3d631a_2
-defaults::readline-7.0-ha6073c6_4
-defaults::setuptools-39.2.0-py36_0
-defaults::sqlite-3.23.1-he433501_0
-defaults::tk-8.6.7-hc745277_3
-defaults::wheel-0.31.1-py36_0
-defaults::xz-5.2.4-h14c3975_4
-defaults::zlib-1.2.11-ha838bed_2
+conda-forge::bzip2-1.0.8-h516909a_0
+conda-forge::ca-certificates-2019.6.16-hecc5488_0
+conda-forge::certifi-2019.6.16-py37_1
+conda-forge::libblas-3.8.0-12_openblas
+conda-forge::libcblas-3.8.0-12_openblas
+conda-forge::libffi-3.2.1-hfc679d8_5
+conda-forge::liblapack-3.8.0-12_openblas
+conda-forge::libopenblas-0.3.7-h6e990d7_1
+conda-forge::ncurses-6.1-hfc679d8_2
+conda-forge::numpy-1.17.0-py37h95a1406_0
+conda-forge::openssl-1.0.2r-h14c3975_0
+conda-forge::pip-19.2.3-py37_0
+conda-forge::python-3.7.1-h5001a0f_0
+conda-forge::readline-7.0-hf8c457e_1001
+conda-forge::setuptools-41.2.0-py37_0
+conda-forge::sqlite-3.28.0-h8b20d00_0
+conda-forge::tk-8.6.9-hed695b0_1002
+conda-forge::wheel-0.33.6-py37_0
+conda-forge::xz-5.2.4-h14c3975_1001
+conda-forge::zlib-1.2.11-h516909a_1005
+defaults::_libgcc_mutex-0.1-main
+defaults::libgcc-ng-9.1.0-hdf63c60_0
+defaults::libgfortran-ng-7.3.0-hdf63c60_0
# update specs: ['numpy']

Because in TLJH we're using conda 4.5.8, I think whatever is going on with the version in tests is not the case in TLJH (the test passes if we use the same conda 4.5.8)

[yuvipanda]

w00t, thanks @GeorgianaElena!