feat(cli): run pipeline from CLI directly

[tumido]

Follow up to: #202 (comment)

@MichaelClifford, @Shreyanand can you please help me understand the default values we set here and if ALL of them make sense in a generic environment. I've made a check-list in code via FIXME: comments. Be aware that the default values are not necessarily expected to work in the MOC cluster/ilab namespace context. (Removed from the PR, we can revisit this separatelly.)

This PR allows us to submit the pipeline run directly via CLI with no drag-and-drop-in-UI-slavery anymore...

All params are optional but allows you to control things:

$ python pipeline.py run --help
Usage: pipeline.py run [OPTIONS]

  Run the pipeline immediately against current kubernetes context (cluster and
  namespace).

  Command sets expected dev-cluster friendly default values when submitting.

Options:
  --mock [sdg|train|eval]  Mock part of the pipeline
  -e, --experiment TEXT    Set KFP experiment name.
  -r, --run TEXT           Set KFP run name.
  -p, --param TEXT         Override default parameters in KEY=VALUE format.
                           Default parameters are suitable for dev cluster -
                           the MOC cluster, `ilab` namespace.
  --help                   Show this message and exit.

Submitting a run then looks like this:

$ python pipeline.py run
Experiment details: https://ds-pipeline-dspa-ilab.apps.ocp-beta-test.nerc.mghpcc.org/#/experiments/details/f3fb689f-8f38-4822-9b6c-aa1c51d8d121
Run details: https://ds-pipeline-dspa-ilab.apps.ocp-beta-test.nerc.mghpcc.org/#/runs/details/2e5bb117-3a3d-44ed-bd11-49864f16df45

[leseb]

How is that different from using the kfp CLI?

[tumido]

@leseb

How is that different from using the kfp CLI?

CLI will become usable against OCP once kubeflow/pipelines#11400 get's into kfp release. Before that there's no way to authenticate KFP CLI properly. And since we're locked on 2.9 at the moment, we can't benefit from it any time soon.