Make registry-backed batch endpoint notebook self-bootstrap its pipeline component in CI

[Copilot]

The from-registry batch endpoint notebook was executing in GitHub Actions but failing when it tried to resolve a registry component using unresolved placeholder values for component_name and component_version. This change makes the sample self-sufficient in CI by ensuring a real pipeline component exists in the target registry before the notebook fetches it.

• Problem

  • CI templated registry_name, but left:
    • <PIPELINE_COMPONENT_NAME>
    • <PIPELINE_COMPONENT_VERSION>
  • The notebook then called registry_ml_client.components.get(...) against a non-existent registry asset and failed at runtime.

• Change

  • Added a conditional bootstrap path in sdk/python/endpoints/batch/deploy-pipelines/from-registry/sdk-deploy-and-test.ipynb.
  • When the notebook still sees placeholder component values, it now:
    • loads the existing reusable pipeline component from the sibling hello-batch sample
    • registers that component into the target registry
    • captures the real returned component name/version
    • retrieves the component from the registry and continues with deployment

• Repo consistency

  • Reuses an existing repo-owned sample component instead of introducing a one-off asset or relying on manual preregistration.
  • Preserves the existing sample flow for real user-provided component coordinates; the bootstrap path only activates for placeholder-based execution such as CI.

• Effect

  • The notebook can now run end-to-end in CI without depending on a pre-seeded registry component.

if component_name.startswith("<") or component_version.startswith("<"):
    pipeline_component = load_component(
        source=Path("../hello-batch/hello-component/hello.yml")
    )
    pipeline_component = registry_ml_client.components.create_or_update(
        pipeline_component
    )
    component_name = pipeline_component.name
    component_version = pipeline_component.version

pipeline_component = registry_ml_client.components.get(
    name=component_name, version=component_version
)