fix(run-queue): handle worker spawn failure gracefully and avoid raw …

[Adityakk9031]

close #38

Description

This PR fixes a critical bug where worker subprocess spawn failures (e.g. OSError raised by subprocess.Popen) during manual lifecycle operations (init, ingest, garden) propagate as raw uncaught Python tracebacks.

We now catch spawn errors safely, wrap them in a product-level ExecutionFailed exception, and present a clean user-facing error message with attach instructions.

Detailed Changes:

• Run Queue Service: Added a safe helper method _spawn_worker_safely in src/codealmanac/workflows/run_queue/service.py to wrap the spawn_worker calls. If spawning fails, it raises an ExecutionFailed exception containing the orphaned run_id and recovery instructions (codealmanac jobs attach <id>).
• Code Size Optimization: Using the helper method avoids adding duplicate try/except blocks across all three triggers, successfully keeping the service line count under the strict architecture test limit of 200 lines.
• Unit Testing: Added test_run_queue_start_handles_worker_spawn_failure in tests/test_run_queue_workflow.py to assert that spawn failures raise ExecutionFailed and the run is kept persisted as QUEUED in the database.

Verification

Automated Tests

Ran the pytest suite for architecture invariants and run queue workflows:

.venv/Scripts/pytest tests/test_architecture.py
.venv/Scripts/pytest tests/test_run_queue_workflow.py

[Adityakk9031]

@divitsheth have a look