Add generated docs for dictionary repos

[snomos]

Use GH Actions to generate info about:

• number of entries
  • possibly also split on POS
• number of translations
• number of examples
• authors

Maybe more.

[Phaqui]

Should it run on every commit? If there are errors in the xml sources, should the script try its best, or write an error?
Where should the information be stored? In a stats.json file in the root, for example?

[Phaqui]

Okay, from the way I understand this now, it seems like it should be possible to commit to the repository from a github action on push to the repositories. However, that new commit is just that, a new commit. This means that users will have to pull after pushing, to retrieve the newly created commit (that the github action commited).

It's possible to run the stats generation script on a calendar interval, e.g. every midnight instead, to cut down on the number of commits. Of course, that may introduce a little bit of an inconsistency in the stats.

I wrote a script in python (and also rust, to compare) which will count lemmas by POS, but we may want to discuss details of this further, before I add such an action.

[Phaqui]

I pushed the script to do this. It is giellalt/giella-core/dicts/scripts/gt_pos_counts.(py|rs).

[snomos]

Thanks!

If we publish using a separate gh_pages branch (as is common), we can build the data as part of the CI process, and it will be committed to that branch only. That is all we need, and that is what we do right now for the maturity metadata in the lang-xxx repos.