Kamu

Added
- Event-sourced implementation of the basic but functional task system
Changed
- Upgraded to latest `datafusion`

Added
- Task system prototyping (in-memory backend + GQL API)
Changed
- More tracing instrumentation for Query service, and ObjectStore builders
Fixed
- Caught root cause of platform's S3 data querying issues

Added
- New `kamu --trace` flag will record the execution of the program and open [Perfetto UI](https://perfetto.dev/) in a browser, allowing to easily analyze async code execution and task performance. Perfetto support is still in early stages and needs more work to correctly display the concurrent tasks.
- Added a few common command aliases: `ls -> list`, `rm -> delete`, `mv -> rename`

Fixed
- Handle "Interrupted system call" error when waiting for container to spawn.

Changed
- Upgraded to new `spark` engine image that better follows the ODF spec regarding the timestamp formats. This may cause schema harmonization issues when querying the history of old datasets.
Added
- Experimental support for new [`datafusion` engine](https://github.com/kamu-data/kamu-engine-datafusion) based on [Apache Arrow DataFusion](https://github.com/apache/arrow-datafusion). Although it's a batch-oriented engine it can provide a massive performance boost for simple filter/map operations. See the [updated documentation](https://docs.kamu.dev/cli/supported-engines/) for details and current limitations.

Fixed
- AWS S3 object store is getting proper credentials from already resolved AWS SDK cache