Fix zip destination path in Hugging Face repo when using custom data directory location

[juhoinkinen]

When using a nondefault data directory location (with ANNIF_DATADIR environment variable) and running annif upload, the zipped projects and vocabs data files end up in a wrong path in the destination repository; the full local data directory gets created to the repo. For example, when using a temp directory, a project datafile with a path like this appears in the repo: tmp/tmp3ttl0hbh/data/projects/tfidf-fi.zip

This is fixed by getting the path to be used in the repo by reading the local datadir path from the right and using two parts of it (the project-id directory and the directory containing it, either projects or vocabs).

There was a unit test for this, but it was asserting a wrong path to be created in the repo.

I noticed this when working with #907.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.63%. Comparing base (371d1eb) to head (5bdb924).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #908   +/-   ##
=======================================
  Coverage   99.63%   99.63%           
=======================================
  Files         103      103           
  Lines        8236     8236           
=======================================
  Hits         8206     8206           
  Misses         30       30           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull Request Overview

This PR updates the path construction logic for uploading project and vocabulary data to Hugging Face Hub to handle nested base data directories correctly.

• Changed path extraction from split(os.path.sep, 1)[1] to rsplit(os.path.sep, 2)[1:3] to reliably extract the last two path components (typename/identifier)
• Updated test assertions to match the new expected paths without the "data/" prefix

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
annif/hfh_util.py	Modified _prepare_datadir_commit to use rsplit for extracting the last two path components, ensuring correct paths regardless of base datadir depth
tests/test_cli.py	Updated test expectations to verify paths are "vocabs/dummy.zip" and "projects/dummy-fi.zip" instead of "data/vocabs/dummy.zip" and "data/projects/dummy-fi.zip"

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[juhoinkinen]

Now I noticed this was only a partial fix: the member paths inside the zip still include the full datadir part:

unzip -l ~/Downloads/tfidf-fi.zip 
Archive:  /home/jmminkin/Downloads/tfidf-fi.zip
Archived by Annif 1.5.0.dev0
  Length      Date    Time    Name
---------  ---------- -----   ----
   406760  2025-11-11 13:45   tmp/data/projects/tfidf-fi/vectorizer
   159508  2025-11-11 13:45   tmp/data/projects/tfidf-fi/tfidf-matrix.npz