feat: Add Chronon online and offline store integrations

[franciscojavierarceo]

Summary

This adds an in-tree optional Chronon integration to Feast and wires a Chronon-specific CI path for it.

The initial integration includes:

• ChrononOnlineStore for online reads through the Chronon feature service
• ChrononOfflineStore and ChrononSource for Chronon-backed offline retrieval from materialized outputs
• ChrononProvider registration so Chronon can be selected as a Feast provider/store type
• unit and integration coverage for the Chronon source/store implementations
• a live-service Chronon integration test that exercises Feast against a real local Chronon service
• a PR workflow that builds Chronon quickstart/service artifacts, boots Mongo + Chronon, and runs the live-service Feast test

Notable details

A couple of issues only showed up once this was tested against the real Chronon service:

• the Feast Chronon client needed to use Chronon's actual HTTP prefix: /v1/features/...
• join/groupby names need to be fully URL-encoded, so quickstart/training_set.v2 becomes quickstart%2Ftraining_set.v2

Those fixes are included here.

Testing

Ran locally:

• uv run pytest -c sdk/python/pytest.ini sdk/python/tests/unit/infra/online_stores/chronon_online_store/test_chronon_online_store.py -q
• uv run pytest -c sdk/python/pytest.ini sdk/python/tests/unit/infra/offline_stores/contrib/chronon_offline_store/test_chronon_source.py -q
• uv run pytest -c sdk/python/pytest.ini sdk/python/tests/integration/online_store/test_chronon_online_store.py sdk/python/tests/integration/offline_store/test_chronon_offline_store.py --integration -q
• JAVA_BIN=/opt/homebrew/opt/openjdk/bin/java infra/scripts/chronon/start-local-chronon-service.sh
• CHRONON_SERVICE_URL=http://127.0.0.1:9000 uv run pytest -c sdk/python/pytest.ini sdk/python/tests/integration/online_store/test_chronon_online_store_real_service.py --integration -q
• infra/scripts/chronon/stop-local-chronon-service.sh

Follow-ups

Open questions for follow-up PRs:

• whether ChrononProvider should stay as a thin passthrough or take on materialization/job orchestration
• whether the Chronon bootstrap should be optimized in CI with artifact caching or a prebuilt image instead of source builds
• how much of the offline path should stay constrained to Chronon materialized outputs vs. broader historical retrieval support