Trajectory RL launches debugging scenario for astropy

A new hard-difficulty debugging scenario, swe-bench-astropy-2, lands in trajrl-bench within 24 hours. Agents must navigate real astropy source code pinned to a bug-present commit and produce a unified-diff patch to make the QDP parser case-insensitive. The scenario includes five sub-tests for partial credit and takes effect at epoch 1464.

• •First debugging scenario in trajrl-bench with real source navigation
• •Agents produce unified-diff patches for partial credit across 5 sub-tests
• •Scenario activates at epoch 1464