[Track] enable e2e process to add new training machine

[sunya-ch]

What would you like to be added?

This is a broken down issue from #1906 to focus on the first task.
We must have an automation ready for new machine to be integrated into Kepler-metal-ci training and validation report.
In addition, we need to prepare a dispatch CI to push the trained model to kepler-model-db to make it available there and also to kepler as well.

Previous issue: sustainable-computing-io/kepler-model-server#258

Why is this needed?

• [Track] Call for Machine/Device Sponsorship #1906
• [Discussion and action required] ec2 data collection workflows (need links to Kepler) kepler-model-server#258
• Add workflow to notify Kepler to update the models once there is a new model release kepler-model-db#24

Action items

• clean up unused workflow files
• generalize the training and validation flow to support different action to create a runner
• add a machine layer for validation result page
• manual workflow to fetch kepler-model-db to add the trained model and push a PR signed by the machine-specific account

Next step (future action items)

• dispatch the PR push workflow on release
• dispatch the PR to kepler repo once the new trained model merged.

[sunya-ch]

During the development of kepler and kepler-model-server, we have to retrain and validate the model for every change to ensure that we have no regression.
The results will report to the github page in the same way we have in https://sustainable-computing-io.github.io/kepler-metal-ci/kepler-model-train-validate.html but per machine. (meaning that no need to upload retrained model during the development)

On each release, we want the model to be exported and be available on the kepler-model-db where other users can use these models.
To export the model, we have the threshold of validation result to allow only accurate model to be pushed and shared.