What

Adds mode and resolve_root controls to the inline Python @task path:

• @task(..., mode=..., resolve_root=...)
• stores the values on CallableRef
• emits them into auto-generated <stem>.components.yaml local_from_python entries
• forwards the values through CallableRef.materialize()
• adds compiler/hydrator tests for sidecar emission and hydrate-time forwarding

This reuses the existing components generate from-python --mode bundle / module_bundler.py machinery; it does not add a new bundler.

Why

Pipeline inline @task components currently auto-emit local_from_python sidecars in default inline mode only. That means generated task runtimes can depend on the container image already shipping first-party Python modules.

For the compile OSS migration, the downstream Discovery Daily Pulse combine-run-data proof needs inline @task wrappers to bundle first-party modules (for example relevance_tools.eval.combine_run_data_core) so the generated component is self-contained with respect to first-party source. External dependencies still come from the image, as they do for standalone bundle mode.

This PR is part 1 of the stack. A downstream Discovery change will bump this submodule commit and switch the combine-run-data inline wrappers to mode="bundle".

Tests

• git diff --check
• uv run ruff check packages/tangle-cli/src/tangle_cli/python_pipeline/task.py packages/tangle-cli/src/tangle_cli/python_pipeline/ref.py packages/tangle-cli/src/tangle_cli/pipeline_compiler.py tests/test_pipeline_compiler.py tests/test_pipelines_cli.py
• uv run pytest tests/test_component_from_func.py tests/test_pipeline_compiler.py tests/test_pipelines_cli.py -q — 253 passed

AI-assisted.