debug

Diagnose distributed-training failures in Halo — hangs/deadlocks, CUDA OOM, loss=NaN/Inf, DeepEP faults, and NCCL collective timeouts — by routing the symptom to the exact opt-in helper, env var, or known failure mode. Auto-fire when a training run hangs/deadlocks, OOMs, produces loss=NaN/Inf, or hits a DeepEP/NCCL fault, or when the user asks to debug, triage, or diagnose such a problem. Grounds every…