Kamu

Changed
- FlightSQL: For expensive queries `GetFlightInfo` we will only prepare schemas and not compute results - this avoids doing double the work just to return `total_records` and `total_bytes` in `FlightInfo` before result is fetched via `DoGet`
- Optimized implementation of Datafusion catalog, scheme, and table providers that includes caching and maximally delays the metadata scanning

Fixed
- FlightSQL: Improved Python connectivity examples (ADBC, Sqlalchemy, DBAPI2, JDBC)
- FlightSQL: Fix invalid `location` info in `FlightInfo` that was causing errors in some client libraries

Fixed
- Bug when handle created during dataset creation had empty account name in a multi-tenant repo.

Added
- Implementation of `ObjectRepository` that can cache small objects on local file system (e.g. to avoid too many calls to S3 repo)
- Optional `S3RegistryCache` component that can cache the list of datasets under an S3 repo to avoid very expensive bucket prefix listing calls

Fixed
- Allow OData adapter to skip fields with unsupported data types instead of chasing

Added
- Experimental support for [OData](https://www.odata.org/) protocol
Fixed
- Pulling datasets by account in multi-tenant workspace