feat: reduce memory usage

[RaczeQ]

• Change GeoDataFrame and DataFrame arguments to special objects ParquetDataTable and GeoDataTable
• Add dataloaders based on streaming from Parquet files
• Rewrite logic in all components (for now just cast to GeoDataFrame / DataFrame)
  • Embedders
    • Count - sinks via polars, maybe rewrite to DuckDB
    • ContextualCount - rewrite to DuckDB?
    • GeoVeX
    • Hex2Vec
    • Gtfs2Vec
    • Highway2Vec
    • S2Vec
  • Joiners
    • IntersectionJoiner
    • IntersectionJoiner - rewrite to DuckDB SedonaDB
    • Implement H3Joiner - Create DGG-based joiners #564
  • Loaders
    • OSM Loaders
      • PBF
      • Online
      • Tiles
      • Highway
    • GTFS Loader
    • Overture Maps Loader
  • Regionalizers
    • Voronoi
    • H3 (rewrite to DuckDB H3, specify if return in int or str index)
    • S2 (rewrite to DuckDB geography)
    • Geocode (maybe?)
    • SlippyMap
    • AdministrativeBoundary (maybe rewrite to OvertureMaps boundaries dataset!)
  • Wrap new objects in plotting module
• Add new IntersectionJoiner based on duckdb-geography extension (with s2 cell covering) or h3 extension (with h3 cell covering).
• Add option to not return geometry in H3Regionalizer or S2Regionalizer
• Change H3 index from str to int to reduce memory usage
• Add option to contextualize any embedding after the fact (export ContextualCountEmbedder logic and create a new function / class)
• Add option to embed any pre-computed count embedding with other classes (additional abstract subclass of CountEmbedder with default transform and transform_count_embeddings that can reuse computed embeddings)

[codecov]

Codecov Report

❌ Patch coverage is 80.32995% with 155 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.12%. Comparing base (7efde62) to head (ece3296).

Files with missing lines	Patch %	Lines
srai/geodatatable.py	86.47%	38 Missing ⚠️
srai/embedders/contextual_count_embedder.py	62.65%	31 Missing ⚠️
srai/loaders/osm_loaders/osm_online_loader.py	34.78%	30 Missing ⚠️
srai/neighbourhoods/h3_neighbourhood.py	51.35%	18 Missing ⚠️
srai/neighbourhoods/_base.py	26.08%	17 Missing ⚠️
srai/loaders/download.py	20.00%	8 Missing ⚠️
srai/h3.py	86.36%	3 Missing ⚠️
srai/embedders/_base.py	75.00%	2 Missing ⚠️
srai/loaders/_base.py	84.61%	2 Missing ⚠️
srai/duckdb.py	95.00%	1 Missing ⚠️
... and 5 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #518      +/-   ##
==========================================
- Coverage   91.59%   89.12%   -2.47%     
==========================================
  Files          63       64       +1     
  Lines        2616     3044     +428     
==========================================
+ Hits         2396     2713     +317     
- Misses        220      331     +111     

Flag	Coverage Δ
ubuntu-latest-python3.10	89.12% <80.32%> (?)
ubuntu-latest-python3.11	89.12% <80.32%> (?)
ubuntu-latest-python3.12	89.06% <80.32%> (-2.53%)	⬇️
ubuntu-latest-python3.9	89.11% <80.32%> (?)
windows-latest-python3.12	89.12% <80.32%> (-2.47%)	⬇️

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

☔ 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.