Mlflow-cratedb

- Update to [MLflow 2.10.2](https://github.com/mlflow/mlflow/releases/tag/v2.10.2)

- Update to [MLflow 2.10.0](https://github.com/mlflow/mlflow/releases/tag/v2.10.0)

- Update to [MLflow 2.9.2](https://github.com/mlflow/mlflow/releases/tag/v2.9.2)
- Update to Python 3.11 across the board

- Update to [MLflow 2.8.1](https://github.com/mlflow/mlflow/releases/tag/v2.8.1)

- Update to [MLflow 2.8.0](https://github.com/mlflow/mlflow/releases/tag/v2.8.0)

Attention: Please note when updating from a previous version:
The database model changed slightly, adding the `storage_location`
column to the `model_versions` table. In order to accommodate the
update, run this SQL DDL command, for example using `crash`.
sql
ALTER TABLE mlflow.model_versions ADD COLUMN storage_location TEXT NULL;

Note that it is always advised to create backups of your database content.
This is an excellent opportunity to do that.

- Fix uniqueness constraints
- `mlflow.server.auth.db.models.SqlUser.username`.
- `m.s.a.d.m.SqlExperimentPermission`: "experiment_id", "user_id"
- `m.s.a.d.m.SqlRegisteredModelPermission`: "name", "user_id"
- Fix OCI build re. `psutil` package on aarch64.
- Optimize OCI image sizes.
- Add example experiment program `tracking_pycaret.py`, and a corresponding
test case. Thanks, andnig.
- Examples: Use `refresh_table` to synchronize CrateDB write operations.

- Fix uniqueness constraint with `SqlRegisteredModel.name`. Thanks, andnig.
- Downgrade to Python 3.10. A few packages like PyCaret are not ready for
Python 3.11 yet. Thanks, andnig.