fix: Update zarr>=3, keep numpy<2 pin, and test notebooks

[Safwannn89]

Hey @LOCEANlloydizard! I went ahead and updated to zarr>=3 and tested the notebooks, but ran into a few interesting things along the way. I initially tried unpinning NumPy completely, but since NumPy 2.x causes netCDF4 to crash with a dtype size error, I put the numpy<2 pin back in for now so the notebooks actually run. I also noticed that the Zarr 3 update seems to break ep.commongrid.compute_MVBS() in get_ship_data; I dug into the traceback and found it failing deep in flox when it tries to index a PyArrow ChunkedArray (throwing a TypeError about integer scalar arrays). The try/except block swallows it so the cell doesn't completely halt, but I want to flag that upstream bug for you. Lastly, the GitHub Actions failed on the Python 3.10 runner because Zarr 3 officially requires Python >=3.11. Since the tests passed perfectly on 3.11 and 3.12, we might just need to drop 3.10 from the workflow entirely. The updated requirements.txt is in this PR, so just let me know how you'd like to handle the CI matrix!

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 69.27%. Comparing base (c9d4bd9) to head (8022d5d).
⚠️ Report is 14 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #242   +/-   ##
=======================================
  Coverage   69.27%   69.27%           
=======================================
  Files          11       11           
  Lines         690      690           
=======================================
  Hits          478      478           
  Misses        212      212           

☔ 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.