refactor: speed up grouping operation

[RaczeQ]

No description provided.

[codecov]

Codecov Report

Patch coverage: 95.65% and project coverage change: -0.16% ⚠️

Comparison is base (48c6c67) 91.74% compared to head (99b82bf) 91.58%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #357      +/-   ##
==========================================
- Coverage   91.74%   91.58%   -0.16%     
==========================================
  Files          60       60              
  Lines        2083     2092       +9     
==========================================
+ Hits         1911     1916       +5     
- Misses        172      176       +4     

Flag	Coverage Δ
macos-latest-python3.11	91.58% <95.65%> (-0.16%)	⬇️
ubuntu-latest-python3.10	91.58% <95.65%> (+0.13%)	⬆️
ubuntu-latest-python3.11	91.58% <95.65%> (-0.16%)	⬇️
ubuntu-latest-python3.8	91.52% <95.65%> (-0.21%)	⬇️
ubuntu-latest-python3.9	91.52% <95.65%> (-0.21%)	⬇️

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

Files Changed	Coverage Δ
srai/loaders/osm_loaders/_base.py	96.29% <95.65%> (-1.49%)	⬇️

... and 1 file with indirect coverage changes

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

[RaczeQ]

Refactored OSMLoader's internal function _group_features_gdf that parsed flat OsmTagsFilter back into GroupedOsmTagsFilter.

The previous version worked directly in a for loop and was slowing down significantly with a higher number of rows in the data frame.

The current version works on whole groups at once (scales linearly with number of groups in filter) and utilizes boolean filter masks and pandas's apply function to work on all rows that match given group in the filter.