Xgboost-ray

* Ranking support (189)
* Fix bad `super()` call in `MLDataset` `DataSource` (184)
* CI fixes (185, 187)

- Preserve column ordering in DMatrices (170)
- Dependency management (171, 173, 174, 175)
- CI improvements (169)
- Documentation/example fixes (178, 180)
- Validate kwargs correctly (172)
- Main: Fix compatibility with XGBoost >= 1.5.0 (179)

- Require up-to-date Ray for compatibility (164)
- Lazily read env variables (166)
- Compatibility with XGBoost >= 1.6.0 (167)

- Add API annotations (158)
- Adjust to Ray API changes (161)
- Fix Dask data loading issues (162)
- Import/dependency fixes, removing `[default]` and adding xgboost as hard dependency again (156, 159)

- Support for more data sources: Ray datasets and general improvements (139, 140, 141, 145)
- Placement group fixes (136, 143)
- Usability improvements (148, 149)
- Various minor fixes/updates (137, 144, 150, 154)
- `xgboost` is now an extra dependency (and might have to be installed separately) (151)

- Improvements in the SKlearn API (125, 126)
- User user warnings instead of logger warnings (128)
- Testing fixes (129)
- Fix backwards compatibility with placement groups (130)
- Improve batching algorithm (132)