[codex] Log SWE rollout group mappings

[rasdani]

Summary

• Log each scheduler rollout group as soon as it is allocated.
• Include step, group_id, example_id, optional instance_id, and env_name so downstream rollout logs can be joined back to scheduler state.
• Add focused unit coverage for the emitted mapping fields.

Validation

• uv run ruff check src/prime_rl/orchestrator/scheduler.py tests/unit/orchestrator/test_scheduler.py
• uv run pytest tests/unit/orchestrator/test_scheduler.py
• unset PRIME_API_KEY; uv run prime-rl --help

A live prime-rl scheduling run was not started because even the smallest end-to-end run needs a trainer/inference cluster configuration and GPU allocation. The scheduler emission is covered by the focused unit test; the companion verifiers branch was validated with a one-row live sandbox rollout.

---

Note

Low Risk
Low risk: adds a new informational log line when a rollout group is created plus a unit test; no scheduling logic or data flow is changed beyond emitting extra logs.

Overview
Adds a scheduler log emitted at rollout-group allocation time so downstream rollout logs can be joined back to scheduler state.

Introduces Scheduler._log_group_created() and calls it when creating a new GroupState, logging step, group_id, example_id, optional instance_id, and env_name; includes a focused unit test asserting those fields appear in the emitted message.

^{Reviewed by Cursor Bugbot for commit bd24148. Bugbot is set up for automated code reviews on this repo. Configure here.}