Add dataloader_multiprocessing_context, default to spawn

Make the DataLoader multiprocessing start method configurable on
TrainPipelineConfig and default it to 'spawn'.

The previous default (fork on Linux) is unsafe with libraries that hold
non-fork-safe state in the parent process — common ones in this codebase
are PyAV, torchcodec, and the ffmpeg shared libs they wrap. Symptoms
reported in #2488, #2209, and observed locally include:

- multiprocessing.context.AuthenticationError: digest received was wrong
- RuntimeError: Pin memory thread exited unexpectedly
- RuntimeError: DataLoader worker exited unexpectedly
- Random SIGSEGV inside worker processes during video decode

Switching to spawn re-imports modules cleanly in each worker and
eliminates these failure modes. Added the setting as a config field
rather than hard-coding so users on platforms where fork is preferred
can opt back in via --dataloader-multiprocessing-context=fork.

src/lerobot/configs/train.py

@@ -101,6 +101,14 @@ class TrainPipelineConfig(HubMixin):
     batch_size: int = 8
     prefetch_factor: int = 4
     persistent_workers: bool = True
+    # DataLoader multiprocessing start method. "spawn" is the safe default on
+    # Linux because workers do not inherit fork-time state from the parent —
+    # "fork" can crash with non-fork-safe libraries that the parent has loaded
+    # (e.g. PyAV / torchcodec / ffmpeg) with errors like
+    # `multiprocessing.context.AuthenticationError: digest received was wrong`,
+    # `Pin memory thread exited unexpectedly`, or random worker segfaults.
+    # See https://github.com/huggingface/lerobot/issues/2488.
+    dataloader_multiprocessing_context: str = "spawn"
     steps: int = 100_000
     # Run policy in the simulation environment every N steps to measure reward/success (0 = disabled).
     env_eval_freq: int = 20_000

src/lerobot/scripts/lerobot_train.py

@@ -475,6 +475,7 @@ def train(cfg: TrainPipelineConfig, accelerator: "Accelerator | None" = None):
         collate_fn=collate_fn,
         prefetch_factor=cfg.prefetch_factor if cfg.num_workers > 0 else None,
         persistent_workers=cfg.persistent_workers and cfg.num_workers > 0,
+        multiprocessing_context=cfg.dataloader_multiprocessing_context if cfg.num_workers > 0 else None,
     )
 
     # Build eval dataloader if a held-out split exists