Edge Feature Quantization [4/5]: Distributed Materialization

[jchmura-sc]

Summary

• Propagates quantized edge features through DistDataset, graph-store, and distributed loaders.
• Materializes reconstructed edge tensors before model consumption.
• Completes the preprocessing-to-loader integration coverage.

Stack

1. Edge Feature Quantization [1/5]: Metadata Proto #749 - metadata proto
2. Edge Feature Quantization [2/5]: Preprocessor #750 - preprocessor
3. Edge Feature Quantization [3/5]: Read and Partition #751 - read path and partitioning
4. Distributed loading and materialization - this PR
5. User guide

Verification

• Ruff and formatting checks pass.
• Python compilation passes.
• The cumulative implementation matches Jchmura/edge feature quantization #745, with its base_dist_loader.py hunks applied on current main to preserve newer PPR behavior.
• Full tests are blocked in this clean worktree because the local environment lacks Beam, Torch, and pytest dependencies.