Feature Request: run inference with a policy from a third-party lerobot plugin #111

lerobot already supports third-party policy plugins — a pip-installed package named lerobot_policy_<name> registers its own policy type into PreTrainedConfig's choice registry (draccus ChoiceRegistry), and lerobot-rollout already calls register_third_party_plugins() at startup to discover and import it. The policy class and its pre/post-processors both resolve through the same dynamic-import-by-naming-convention lerobot already has (configuration_<type>.py, modeling_<type>.py, processor_<type>.py) — nothing needed on lerobot's side, and nothing needed in LeLab's lelab/rollout.py either: _resolve_policy_path() already accepts an arbitrary local checkpoint directory or Hub ref verbatim.

The actual gap is the UI. InferenceModal.tsx only ever builds policy_ref from /jobs/{id}/checkpoints — checkpoints from one of your own LeLab training runs. There's no way to point inference at a checkpoint that isn't tied to a LeLab job: a plugin-trained policy sitting in a local directory, or a Hub repo you didn't train through LeLab.

Fix: add a free-text "custom policy path / Hub ref" input to the inference modal as an alternative to the job-checkpoint dropdown, wired straight to policy_ref the same way. As long as the plugin package (lerobot_policy_<name>) is pip-installed in the same environment LeLab's lerobot-rollout subprocess runs in, and the checkpoint's config.json declares that type, it should already just work.