DO NOT MERGE: Timing test files

[datapythonista]

Checking how long every test file takes to run in the CI. In #26949 I'd like to run tests constraining that a whole file must be run in a single worker (so we can run single tests in parallel).

This means that files that will take a lot of time won't be parallelized well, and should be splitted for this to work. So, I want to know how long each takes.

[datapythonista]

Checking how long it takes every test file to see if some should be splitted for better parallelization I see this, which doesn't look right to me:

pandas/tests/io/test_html.py : 21m59.041s
pandas/tests/io/json/test_pandas.py : 8m50.339s
pandas/tests/io/parser/test_common.py : 6m45.860s
pandas/tests/io/excel/test_readers.py : 13m14.864s

Didn't check the files individually, I guess in some cases we want to test loading big files, but 22 minutes to test the html I/O sounds very exaggerated.

@pandas-dev/pandas-core is this expected?

[jorisvandenbossche]

Did you try running that single file locally?

[datapythonista]

No, wanted to check first that this is not expected and I'm missing something. If this doesn't seem right, I'll then try to figure out what's going on.

[jorisvandenbossche]

The second one takes locally 4.6 seconds (and not almost 9 min). There are 2 out of 97 tests skipped (encoding related), of which one is an s3 test

[jbrockmendel]

pandas/tests/io/test_html.py

I'm seeing the two (not marked as slow) test_invalid_url tests as about 150s apiece.

[jbrockmendel]

@datapythonista closable?

[datapythonista]

Yep, let's close this for now, may need to reopen later to research on the timing problem, but I don't have time now.