feat(sm-integration): Add SageMaker training as a backend.

[Linbo-Liu]

Issue #, if available:

Description of changes:

What

Adds an experimental SageMaker Training Sessions backend: a GRPO loop that trains an AgentCore Runtime (ACR)-deployed agent with no local GPU cluster. SageMaker hosts the policy weights, the sampler, and the optimizer behind an SDK, so the RL loop itself is a plain single-process asyncio program that can run on a laptop or a small EC2 box. Token capture reuses the in-repo rollout gateway, exactly as the experimental verl backend does.

How it fits together

The only new engine seam is SageMakerSdkBackend (rollout_gateway/sampling_backends/sagemaker_sdk.py) — a gateway SamplingBackend over the SageMaker SamplingClient that maps token_ids -> token_ids + logprobs. Like TinkerSdkBackend, it does not render: the gateway owns tokenization, which keeps loss-masking well-defined and matches the existing placement rule (independently reachable hosted SDK → sampling_backends/).

The training loop

backends/experimental/sagemaker/ is driven by a single YAML config (config.py / config.yaml.example):

File	Role
train_grpo.py	GRPO loop: rollout → advantages → forward_backward → optim_step → rebind sampler
rollout.py	One ACR rollout: create gateway session, invoke the agent, await the S3 result, drain trajectories
datum.py	TraceRecord + advantage → SageMaker training datum
config.py / config.yaml.example	Typed config dataclass and a template to copy
prepare_datasets/prepare_gsm8k.py	GSM8K → payload-column Parquet

Tests

Tested on GSM8k. Stable training for 200+ steps. Test accuracy increases from ~70% -> ~90% with gpt-oss-20b.

Docs

New docs/site/src/content/docs/guides/sagemaker-backend-setup.md (installation, config reference, what the loop does step by step, current limits), wired into the sidebar and linked from index.mdx and guides/overview.mdx alongside slime / rllm / verl. A short SETUP.md sits next to the code for readers who get there from the source tree.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.