Post job hallucination/synthetic analysis pipeline/skill

Instead of a single llm call for post analysis job completion cleanup through the orchestrator, have another agentic pipeline that processes the scripts that were generated by the analysis agent.

it should:
- assess the verdict_results.json and run api calls to the backend to update the hypothesis table
- update the profile.md and insights.md with whatever important nuance was observed in the report.md from the job
- potentially auto add hypotheses into the queue based on some confidence/novelty metric

Start by tracing post-job completion cleanup through the orchestrator and inspect the generated verdict_results.json, report.md, profile.md, and insights.md artifacts. Review the backend API and hypothesis-table flow before defining how the pipeline should assess results, preserve report nuance, and decide whether hypotheses enter the queue.

Written by the indexing model from the issue text.