Implementation: Generic GTFS Diff Engine

[cka-y]

Describe the problem

There is no reusable, infrastructure-agnostic Python module for computing the difference between two GTFS datasets. Any diff logic built inside mobility-feed-api risks being tightly coupled to MobilityData-specific concepts (feed IDs, GCS paths, database schemas), preventing reuse across projects or publication as a standalone community tool.

Proposed solution

Create a generic gtfs_diff Python package in a standalone GitHub repository. The package takes two GTFS dataset sources (URLs or local paths) and produces a structured changelog. It must have zero knowledge of MobilityData-specific concepts (no feed IDs, no GCS paths, no database).

Acceptance criteria:

• GtfsDiff.compute() correctly identifies added, removed, and modified rows across all supported GTFS files
• Unit tests cover: all-same feeds (empty diff), one feed empty, added rows, removed rows, modified rows, unsupported file gracefully skipped
• Package is importable locally after pip install -e .
• README.md includes install instructions and a usage example
• Zero references to MobilityData-specific infrastructure in any module

Alternatives considered

• Inline the diff logic inside gtfs-change-tracker: creates tight coupling, prevents reuse as a standalone package.
• Keep the module inside mobility-feed-api: still couples the logic to MobilityData infrastructure and makes PyPI publication awkward.