Fix for HeteroData node indexing in LocalGraphStore and LocalFeatureStore

[JakubPietrakIntel]

To align partitioned data storage in LocalGraphStore and LocalFeatureStore with standard format for PyG HeteroData we need to keep indexing for the nodes of the same type separate, i.e. each node type has indices starting at 0 (rather than using global indexing for all).

[codecov]

Codecov Report

Attention: 15 lines in your changes are missing coverage. Please review.

Comparison is base (408d23e) 87.71% compared to head (b88d096) 87.69%.

❗ Current head b88d096 differs from pull request most recent head 7c2cde8. Consider uploading reports for the commit 7c2cde8 to get more accurate results

Files	Patch %	Lines
torch_geometric/loader/cluster.py	85.71%	7 Missing ⚠️
torch_geometric/distributed/local_graph_store.py	14.28%	6 Missing ⚠️
torch_geometric/distributed/local_feature_store.py	66.66%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8438      +/-   ##
==========================================
- Coverage   87.71%   87.69%   -0.03%     
==========================================
  Files         478      478              
  Lines       29370    29400      +30     
==========================================
+ Hits        25763    25781      +18     
- Misses       3607     3619      +12     

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