Send Prime monitor display config to platform

[d42me]

Updates PrimeMonitor registration so platform external runs receive the resolved run config and display fields needed for correct dashboard values.

PrimeMonitor is constructed by the orchestrator, so registration now treats run_config as an OrchestratorConfig and reads the fields directly — student.model.name for base_model, train.env for environments, group_size for the platform rollouts_per_example field, plus batch_size, seq_len, max_steps, and the full serialized run_config — instead of duck-typed getattr probing over config shapes that do not reach this code path.

Removed the reviewer-requested PrimeMonitor registration unit test case; the remaining Prime monitor unit tests cover rollout parquet serialization and JSON payload sanitization.

Verified:

• uvx ruff check tests/unit/utils/test_prime_monitor.py src/prime_rl/utils/monitor/prime.py

---

Note

^{Cursor Bugbot is generating a summary for commit b3b9d7f. Configure here.}

[d42me]

Updated this PR with:

• removed the requested PrimeMonitor registration unit test case
• fixed external-run environment metadata for RL config shapes where envs live under train.env or orchestrator.train.env

Validation on dev box root@95.133.252.95:

• pytest -q test_prime_monitor.py → 3 passed
• ruff check src/prime_rl/utils/monitor/prime.py test_prime_monitor.py → passed
• prime-rl → platform external-run registration smoke → passed for both orchestrator-shaped and full RLConfig-shaped payloads

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 52b79ea. Configure here.}