Use proper type hint for griffe 0.22 compatibility

[JPBergsma]

Griffe 0.22.0 does not seem to handle type declarations with multiple types in the doc string. So I moved the type declaration to the definition of the function recursive_postprocessing in optimade.filtertransformers.mongo.py
This should allow PR #1261 to be merged with the master branch.

[codecov]

Codecov Report

Merging #1313 (70e177a) into master (88ffc37) will not change coverage.
The diff coverage is 100.00%.

❗ Current head 70e177a differs from pull request most recent head 5196bcb. Consider uploading reports for the commit 5196bcb to get more accurate results

@@           Coverage Diff           @@
##           master    #1313   +/-   ##
=======================================
  Coverage   90.90%   90.90%           
=======================================
  Files          72       72           
  Lines        4364     4364           
=======================================
  Hits         3967     3967           
  Misses        397      397           

Flag	Coverage Δ
project	90.90% <100.00%> (ø)
validator	90.90% <100.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
optimade/filtertransformers/mongo.py	96.66% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[ml-evs]

Could you cherry-pick 69b58ea into this PR so we can test that it does indeed work with the new version? Might be nice to add type hints throughout this submodule too, but not urgent

[ml-evs]

I've just done the cherry-pick, will merge if it passes

[JPBergsma]

Thanks, I got a merge conflict when I tried to cherry-pick it.

[JPBergsma]

Shall I add these changes to #1261 instead ?

[ml-evs]

Shall I add these changes to #1261 instead ?

Nah, I think we just want to stop pinning griffe altogether, if we do that in the dependabot PR it will get confused. Not sure why the docker image test is failing on this though, will see if it works on a restart

[JPBergsma]

It seems as if the commas have been escaped twice. %2C is the escape for a comma in the url and %25 is the escape character for the % sign. No idea why this happened though.

[ml-evs]

Just reran the same tests on the head of master, looks like the docker build is pulling in newer versions of a few packages (see this same failure: https://github.com/Materials-Consortia/optimade-python-tools/runs/8101616258?check_suite_focus=true) (we should consider pinning deps in the Docker image).

I'll trigger dependabot so we can figure out which dep is causing this.

[ml-evs]

Looks like the issue is caused by the latest pydantic (see #1314). I've just added a commit here that uses the pinned dependency versions in the Docker file. As this is built and pushed to the GitHub container registry, we were effectively doing this anyway, so the change only affects those who are building the Docker image from scratch themselves.

[ml-evs]

Looks like the issue is caused by the latest pydantic (see #1314). I've just added a commit here that uses the pinned dependency versions in the Docker file. As this is built and pushed to the GitHub container registry, we were effectively doing this anyway, so the change only affects those who are building the Docker image from scratch themselves.

This worked, I'm going to separate it into another PR.

[ml-evs]

Thanks for this @JPBergsma, and for revealing the other bug too!

[JPBergsma]

A thanks to you too @ml-evs, for figuring out that the pydantic update caused the error message we were seeing.