Kamu

Added
- OData API now supports querying by collection ID/key (e.g. `account/covid.cases(123)`)
Fixed
- Handle broken pipe panic when process piping data into `kamu` exits with an error
- GraphQL Dataset Endpoints object: tenant-insensitive paths & updated REST API push endpoint

Fixed
- Add precondition flow checks
- Fix URLs for Get Data panel

Added
- Added persistence infrastructure prototype based on `sqlx` engine:
- supports Postgres, MySQL/MariaDB, SQLite database targets
- sketched simplistic Accounts domain (not-integrated yet)
- converted Task System domain to use persistent repositories
- added test infrastructure for database-specific features
- automated and documented development flow procedures in database offline/online modes

Added
- Support `ArrowJson` schema output format in QGL API and CLI commands
- New `kamu system compact <dataset>` command that compacts dataslices for the given dataset
Changed
- Case-insensitive comparisons of `dataset`s, `account`s and `repo`s

Added
- Added GraphQL Dataset Endpoints object
Changed
- REST API: `/ingest` endpoint will return HTTP 400 error when data cannot be read correctly
- Improved API token generation command

Changed
- Updated embedded Web UI to `v0.17.0`
Fixed
- S3 Repo: Ignore dataset entries without a valid alias and leave them to be cleaned up by GC
- Caching object repo: Ensure directory exists before writing objects