fix(guard): tighten runtime hook state handling

[capy-ai]

Summary

• tighten post-execution hook state handling
• add regression coverage for runtime artifact routing

Testing

• uv run pytest -q tests/test_guard_runtime.py -k 'routes_post_tool_package_installs_without_pretool_decision or does_not_requeue_package_request_after_pretool_block or routes_package_installs_to_package_request'

[greptile-apps]

Greptile Summary

Tightens runtime hook state handling for post-tool package install checks.
Normalizes camelCase preExecutionResult into the internal pre_execution_result field.
Preserves duplicate suppression when post-tool payloads already carry a pre-tool decision.
Adds regression coverage for package request routing and pre-tool block handling.

Confidence Score: 5/5

The changes are narrowly scoped to runtime hook state handling and are covered by targeted regression tests.

No correctness issues were identified in the modified hook payload/artifact routing paths, and the included tests exercise the relevant package-request and pre-tool decision scenarios.

T-Rex Logs

What T-Rex did

• Before the change, the base did not route the no-pretool PostToolUse npm install to package_request and did not normalize preExecutionResult into pre_execution_result.
• After the change, the head creates a package_request for npm install react@18.3.0.
• The pre_execution_result is normalized to 'block'.
• The system suppresses duplicate artifact creation for the already-blocked install.
• Two targeted pytest regressions were successfully passed.

_{Ran code and verified through T-Rex}

_{Reviews (3): Last reviewed commit: "fix: handle camelcase hook pre-exec deci..." | Re-trigger Greptile}