Add MinIO & MLFlow configuration to notebooks in form of configuration

[Barteus]

We need a way to pass the MinIO & MLFlow configurations to the notebooks (2 separate configuration values in dropdown). Best if similar to the KFP configuration. #49

Expected
Minio configuration in notebook creation screen section Configurations. Additionally added to the documentation.

Workaround 1
Add cell in notebook:

os.environ['MLFLOW_TRACKING_URI'] = "http://mlflow-server.kubeflow.svc.cluster.local:5000"
os.environ["MLFLOW_S3_ENDPOINT_URL"] = "http://minio.kubeflow.svc.cluster.local:9000"
os.environ["AWS_ACCESS_KEY_ID"] = "minio"
os.environ["AWS_SECRET_ACCESS_KEY"] = "minio123"

Workaround 2
Add Pod Defaults with the above configuration to the user namespace, which will be shown in the jupyter notebooks creation screen.

[ca-scribner]

If we solve this via PodDefault, this depends on us having a way to inject those PodDefault objects into user namespaces (see jira tasks KF-220/KF-228)

[ca-scribner]

A potential issue with this is that if the PodDefaults directly inject credentials, those credentials would be in the pod's spec as environment variable literals (eg: you can read the environment variable's content from kubectl describe pod ___). To obscure this, I think we'd need PodDefaults that inject environment variables as secret references (similar to how we pass creds to our workload pods like in Minio), and then we'd also need a way of injecting Secrets with the actual creds into each namespace.