🐛 fix v30_to_v21 ArrowTypeError on pandas extension dtypes

`table.slice(...).to_pandas()` produces pandas ExtensionArrays for
`array[float32]` columns (e.g. `observation.states.end.orientation`)
on newer pandas/pyarrow combos, which then fail in
`pa.Table.from_pandas` inside `Dataset.from_pandas(...).to_parquet(...)`.

Skip the pandas round-trip and wrap the `pa.Table` slice in a
`Dataset` directly with `Dataset(episode_table).to_parquet(...)`.
This preserves the HuggingFace dataset metadata that `Dataset.to_parquet`
writes, while avoiding the ExtensionArray crash. No version pin on
datasets/pyarrow needed.

Closes #87

ds_version_convert/v30_to_v21/convert_dataset_v30_to_v21.py

@@ -181,7 +181,7 @@ def convert_data(root: Path, new_root: Path, episode_records: list[dict[str, Any
                     f"episode_index={episode_index}, length={length}"
                 )
 
-            episode_table = table.slice(start, length).to_pandas()
+            episode_table = table.slice(start, length)
 
             dest_chunk = episode_index // DEFAULT_CHUNK_SIZE
             dest_path = new_root / LEGACY_DATA_PATH_TEMPLATE.format(
@@ -189,7 +189,7 @@ def convert_data(root: Path, new_root: Path, episode_records: list[dict[str, Any
                 episode_index=episode_index,
             )
             dest_path.parent.mkdir(parents=True, exist_ok=True)
-            Dataset.from_pandas(episode_table).to_parquet(dest_path)
+            Dataset(episode_table).to_parquet(dest_path)
 
 
 def _group_episodes_by_video_file(