It seems that using Zenodo as a backup data storage was not as smart as I thought it would be. Seems to be heavily throttled so after 4 successful fetches from individual workers within a short time, the rest is blocked.

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.4%. Comparing base (e47859e) to head (22bd9fd).
⚠️ Report is 9 commits behind head on main.

@@          Coverage Diff          @@
##            main    #818   +/-   ##
=====================================
  Coverage   85.4%   85.4%           
=====================================
  Files        150     150           
  Lines      15954   15962    +8     
=====================================
+ Hits       13620   13629    +9     
+ Misses      2334    2333    -1     

I cut a new release of geodatasets linking to Github url instead of Zenodo. That seems to solve the issue. In any case, we should pre-fetch the data.

I also pushed fixes to rest of the tests, so we're green.