fix: split bootstrap registry and stabilize CrewAI test

[sipercai]

Description

This PR fixes the CrewAI CI failure reported in #230 and reduces recurring LoongSuite bootstrap registration conflicts.

Changes included:

• make the CrewAI authentication-failure test use non-empty fake provider keys so the provider validation reaches kickoff deterministically without falling through to an unrelated live OpenAI API call;
• split loongsuite-distro bootstrap registrations into package-local bootstrap_registry/*.py fragments while preserving the existing bootstrap_gen.py exports;
• update the LoongSuite bootstrap generator and add generator tests for registry fragments;
• update the dynamic LoongSuite CI detector so LoongSuite registry fragment-only changes can scope to the owning package, while loader/upstream/shared registry changes still trigger full runs.

Fixes #230

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

• python "$PIPELINE_SKILL_DIR/scripts/check_loongsuite_pr_readiness.py" --repo .
• tox -c tox-loongsuite.ini -e py312-test-detect-loongsuite-changes
• tox -c tox-loongsuite.ini -e py312-test-loongsuite-instrumentation-crewai -- instrumentation-loongsuite/loongsuite-instrumentation-crewai/tests/test_error_scenarios.py::TestErrorScenarios::test_api_key_missing
• python -m build loongsuite-distro --wheel --outdir .tmp-wheel-test
• PRE_COMMIT_HOME="$(mktemp -d /tmp/loongsuite-precommit.XXXXXX)" tox -e precommit

Does This PR Require a Core Repo Change?

• Yes. - Link to PR:
• No.

Checklist:

See contributing.md for styleguide, changelog guidelines, and more.

• Followed the style guidelines of this project
• Changelogs have been updated
• Unit tests have been added
• Documentation has been updated

Validation Evidence

Spec and Scope

• Linked issue/spec: Fixes ci: crewai instrumentation tests failing on main (VCR cassette + OpenAI API connection error) #230.
• Approved spec/comment: direct maintenance fix requested for issue ci: crewai instrumentation tests failing on main (VCR cassette + OpenAI API connection error) #230 plus the bootstrap registration conflict/CI full-run problem.
• Changed surface: CrewAI test error path, loongsuite-distro bootstrap registry generation/loading, and dynamic LoongSuite CI change detection.

Local Checks

Check	Command	Result	Notes
Static readiness	python "$PIPELINE_SKILL_DIR/scripts/check_loongsuite_pr_readiness.py" --repo .	pass	LoongSuite static PR readiness checks passed.
Precommit	PRE_COMMIT_HOME="$(mktemp -d /tmp/loongsuite-precommit.XXXXXX)" tox -e precommit	pass	Fresh hook cache; ruff, ruff format, uv-lock, and rstcheck passed.
Focused tests	tox -c tox-loongsuite.ini -e py312-test-detect-loongsuite-changes	pass	45 detector/generator tests passed.
Focused tests	tox -c tox-loongsuite.ini -e py312-test-loongsuite-instrumentation-crewai -- instrumentation-loongsuite/loongsuite-instrumentation-crewai/tests/test_error_scenarios.py::TestErrorScenarios::test_api_key_missing	pass	The tox command ran the full CrewAI test directory; 48 tests passed.
Package build	python -m build loongsuite-distro --wheel --outdir .tmp-wheel-test	pass	Wheel contained bootstrap_registry/__init__.py and 81 registry Python files.
Bootstrap compatibility	compare current registry output with origin/main bootstrap_gen.py	pass	libraries and default_instrumentations matched exactly, including order.
External review	external review gate	pass	One review round, 0 findings.
Privacy scan	`git diff --name-only origin/main...HEAD | rg 'tests/cassettes/	.ya?ml$	.json$'`

Real E2E Matrix

Scenario	Status	Command or Demo	Evidence
non-streaming	pass	tox -c tox-loongsuite.ini -e py312-test-loongsuite-instrumentation-crewai -- instrumentation-loongsuite/loongsuite-instrumentation-crewai/tests/test_error_scenarios.py::TestErrorScenarios::test_api_key_missing	Full CrewAI test directory ran, including sync LLM and agent workflow coverage.
streaming	pass	same CrewAI tox command	Full CrewAI test directory ran, including streaming LLM tests.
concurrency	N/A	N/A	This PR does not change CrewAI runtime context propagation or concurrent execution behavior.
agent/tool/ReAct	pass	same CrewAI tox command	Full CrewAI test directory ran, including agent workflow and tool call tests.
tool-heavy	pass	same CrewAI tox command	Full CrewAI test directory ran, including multiple-tool coverage.
error path	pass	same CrewAI tox command	Authentication, sync error, streaming error, and tool error tests passed.

Telemetry and Weaver

Check	Status	Command or Artifact	Notes
Span tree / span kinds	pass	CrewAI focused tox suite	Existing CrewAI tests validate emitted spans for the exercised workflows; this PR does not change span semantics.
Content capture modes	pass	CrewAI focused tox suite	Existing content-capture tests passed as part of the full CrewAI tox run.
Concurrency isolation	N/A	N/A	No runtime concurrency or context propagation logic changed.
Weaver live-check	N/A	N/A	This PR changes test determinism and bootstrap registration metadata, not GenAI semantic attributes or span schema.

CI

• GitHub checks: pending; this PR should trigger CI after publication.
• Known unrelated failures: none verified on this branch after publication.
• Follow-up needed: inspect GitHub checks after they start.

[sipercai]

Superseded by #233, which integrates this bootstrap registry and CrewAI stabilization work together with #208 and #229 and includes the combined validation evidence. Closing this PR to keep review on the integrated branch.