Log SFT metrics during training

[Kovbo]

Summary

• Log SFT optimizer metrics at configurable gradient-step intervals without creating per-step checkpoints. This adds a separate sft/* W&B namespace for detailed SFT metrics, using sft/gradient_step as the x-axis. SFT jobs still produce one checkpoint and one aggregate training metric, while detailed per-step metrics are logged under the separate SFT split.
• Keep local backend logging client-owned, while letting remote SFT backends own server-side metric logging. This is intentionally different from RL: for Serverless SFT, the client may disconnect after starting training, so server-side logging is required if we want metrics to continue being recorded.
• Forward SFT metric logging config through serverless training jobs.
• Define W&B routing for the sft/* namespace.
• Update dev SFT scripts to exercise Megatron/Qwen SFT.

Testing

• uv run ty check src tests
• uv run pytest tests/unit/test_frontend_logging.py::TestTrainSFTMetricsAggregation tests/unit/test_metric_routing.py tests/unit/test_serverless_pipeline_trainer_compat.py -q
• uv run python -m py_compile dev/sft/sft-from-file.py dev/sft/sft-warmup.py