feat: Add OnlineStore for MongoDB

[caseyclements]

What this PR does / why we need it:

Adds a first-class MongoDB online store integration at feast.infra.online_stores.mongodb_online_store.

Schema

Each entity is stored as a single MongoDB document keyed by its serialized entity key:

{
  "_id": "<serialized_entity_key>",
  "features": { "<feature_view>": { "<feature>": <value> } },
  "event_timestamps": { "<feature_view>": "<datetime>" },
  "created_timestamp": "<datetime>"
}

Because MongoDB has a loose schema and supports upsert semantics natively, update() requires no
pre-creation — it only removes feature views named in tables_to_delete via $unset. Multiple
feature views for the same entity share one document.

Implementation highlights

• Sync and async paths — implements online_write_batch, online_read,
  online_write_batch_async, online_read_async, and async def close().
  async_supported returns read=True, write=True.
• requested_features projection — both read paths build a MongoDB field projection so only
  the requested feature columns are returned from the server.
• Efficient BSON → proto conversion — _convert_raw_docs_to_proto transforms column-wise to
  minimise calls to python_values_to_proto_values (one call per feature across all entities,
  rather than one call per entity × feature). Benchmarking confirmed this is ~3–4× faster than a
  naïve row-wise approach across all scaling dimensions (entities, features, feature views).
• Composite entity keys — entities are serialized with serialize_entity_key, so composite
  keys (e.g. customer_id + driver_id) are handled without schema changes.

Tests

• Docker integration test (test_mongodb_online_features) — spins up a real MongoDB container
  via testcontainers, writes to three feature views (single int key, single string key, composite
  key), and asserts correct retrieval including type coercion and missing-entity handling.
• Pure Python unit tests (no Docker required):
  • test_convert_raw_docs_missing_entity — entity absent from query results → (None, None)
  • test_convert_raw_docs_partial_doc — entity present but a feature key missing → empty
    ValueProto (schema-migration safety)
  • test_convert_raw_docs_ordering — result order follows the requested ids list regardless
    of MongoDB cursor order

All tests pass. Code is clean under mypy, ruff check, and ruff format.

---

[shuchu]

@caseyclements please check the errors in the tests :)