Address slow/flaky tests

[khaeru]

One or more of the following tests appear to often stall on GitHub Actions:

message_ix_models.tests.model.material.test_build.test_build_bare_res[R12-B-B-False]
message_ix_models.tests.transport.report.test_simulated_iamc
message_ix_models.tests.util.test_common.TestWildcardAdapter.test_0

To resolve:

• Identify the causes.
• Adjust the test or the underlying code.

[khaeru]

Initially this issue was just about .transport.report.test_simulated_iamc(). However, from inspecting the job logs (details below), it seems there are cases where that test passes, and instead one or both of the following tests may be responsible for the timeout:

tests/model/material/test_build.py::test_build_bare_res[R12-B-B-False]
tests/util/test_common.py::TestWildcardAdapter::test_0

So I've updated the description and title.

---

These are identified using the raw job logs, which contain lines like the following:

message_ix_models/tests/util/test_slurm.py::test_cli[sbatch_opts0-env0] 
[gw1] [ 97%] PASSED message_ix_models/tests/util/test_slurm.py::test_cli[sbatch_opts0-env0] 

In these:

• Line 1 is an indication that the particular test is being dispatched to a worker using pytest-xdist
• Line 2 is an indication from the particular worker that the test has completed, along with its status.

By search-and-replace, the second line can be edited:

message_ix_models/tests/util/test_slurm.py::test_cli[sbatch_opts0-env0] 
message_ix_models/tests/util/test_slurm.py::test_cli[sbatch_opts0-env0] 

…and then the whole file run through cat job-logs.txt | sort | uniq -c, which gives a count of the number of times each line occurs:

      2 message_ix_models/tests/util/test_slurm.py::test_cli[sbatch_opts0-env0] 

Wherever the initial digit is 1 (not 2), it indicates that the test was dispatched to a worker, but never returned before the log/job ended, i.e. may have been the culprit for the timeout.

[khaeru]

FYI @glatterf42 @macflo8.

I can remain the assignee for the issue overall, but if I end up touching part of the material test I would ask @macflo8 for confirmation that I am not disturbing anything.