[Feature] Add mlflow experiment tracker

[debrevitatevitae]

Modify training routines and config to be able to log with ML-Flow.

TODOs:

• Add MLConfig with username/pw fields which can be set using os.getenv
• Add StrEnum EXPERIMENT_TRACKER with options TENSORBOARD and MLFLOW
• Add EXPERIMENT_TRACKER to TrainConfig as tracking_tool
• Modify train loop to either perform write_tensorboard or write_mlflow depending on the setting of tracking_tool is
• Add log_params for configs
• Add log_figures using callbacks
• Add log_model (overlap with checkpointing?)
• Test integration with cloud

TODOS @debrevitatevitae:

• FInalize interface/ features for mlflow tracking
• Align train_no_grad with train_grad
• Create a tutorial showcasing mlflow with remote tracking
• Add tests with mlflow tracking

[Roland-djee]

@dominikandreasseitz @debrevitatevitae Is there review needed here ? Sorry, I lost track a little.

[Roland-djee]

Thanks @debrevitatevitae some cosmetic changes from me.

[dominikandreasseitz]

@debrevitatevitae please try to finish the PR with @Roland-djee's help thanks!

[Roland-djee]

Thanks a lot @debrevitatevitae, that's really nice and clean. Some minor comments from me. Happy to approve once addressed.