Reorganise pipeline YAML config

Historically, the `kernelci-core` repository had its configuration for builds and boot tests directly built-in as Python dictionaries. Then this got converted to YAML but stayed in the same repository with the code. With the new API & Pipeline, the `kernelci` Python package should be decoupled from the YAML configuration which is really specific to each pipeline instance and client-side services.

It probably makes sense to provide some standard entries such as the public API instances, upstream Git kernel trees, local runtimes (shell, docker) and some job definitions with a basic scheduler. This could be installed in `/etc/kernelci` as is currently the case in the `kernelci/kernelci` Docker image, then users could supply additional files in a local directory with their own specific entries as an overlay. For the production KernelCI instance, these files would be stored in `kernelci-pipeline` as per the current proof-of-concept `pipeline.yaml` file there. The same principle can apply with Jinja2 templates and other configuration files, with a same directory structure as in `/etc/kernelci` but in a local user directory.

- [ ] Investigate whether this would provide a long-term solution to deploy YAML pipeline configuration
- [ ] Plan deprecation for legacy YAML files and related code
- [ ] New proposal implemented
- [ ] Tested on staging with new API
- [ ] Verified that legacy code is not impacted
- [ ] Merged and ready for production deployment

Review the kernelci Python package, the configuration bundled in the kernelci Docker image, and the proof-of-concept pipeline.yaml in kernelci-pipeline. First compare the current YAML layout with the proposed /etc/kernelci and local overlay structure. Done means a documented proposal, a legacy deprecation plan, staging validation with the new API, and confirmation that legacy code remains unaffected.

Written by the indexing model from the issue text.