feat: add FSDP2Executor using torch.distributed.fsdp.fully_shard API

- New FSDP2Executor registers as 'fsdp2' in ExecutorFactory, using per-module fully_shard() instead of FSDP1 FlatParameter wrapper
- FSDP2 preserves original Parameter objects as DTensors, eliminating use_orig_params=True hack
- FSDP2Executor implements _no_sync via set_requires_gradient_sync, clip_grad_norm via unshard, unwrap_model via DTensor.full_tensor
- Drop **_extra/**_ddp_only_kwargs fallbacks in BaseExecutor/FSDPExecutor/FSDP2Executor, replaced by parallel_mode-aware executor_kwargs dispatch in train.py (ddp-only kwargs only passed when parallel_mode=ddp)
- Export FSDP2Executor in astrai.parallel.__init__

scripts/tools/train.py

@@ -449,10 +449,12 @@ def train(
         "group_size": kwargs.pop("group_size"),
     }
 
-    executor_kwargs = {
-        "gradient_as_bucket_view": True,
-        "broadcast_buffers": False,
-    }
+    executor_kwargs = {}
+    if parallel_mode == "ddp":
+        executor_kwargs.update(
+            gradient_as_bucket_view=True,
+            broadcast_buffers=False,
+        )
 
     model_fn = partial(create_model, config)
     dataset = DatasetFactory.load(