LLM-chain: database schema for chain execution tracking

[vprashrex]

Is your feature request related to a problem? Please describe.
Currently, the job table only tracks overall chain status (PENDING → PROCESSING → SUCCESS/FAILED). Individual block executions are stored as separate llm_call records but there's no consolidated view of which block is running, which completed, or where a chain failed. If block 3 of 5 fails, we only see "FAILED" on the job, no checkpoint info to debug or resume from.

Describe the solution you'd like
Add chain-level execution tracking that stores per-block checkpoint state (block index, status, input, output, error) so the entire chain's progress is visible from a single query or best would be in single rwo.

Describe alternatives you've considered
Reference how Airflow tracks per-task state within a DAG Run (TaskInstance table) and how Temporal stores workflow execution history with per-activity checkpoints.

Additional context
Schema should persist that block-level state to the database for queryability and potential resume support.