Default hdfs.replication to 3 (match PyArrow default)

Keep PyArrow's original default behavior. Users can override
per-table or per-catalog via hdfs.replication property.
Set to 0 to defer to server-side hdfs-site.xml.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Author: 2 people

pyiceberg/io/pyarrow.py

@@ -579,7 +579,7 @@ def _initialize_hdfs_fs(self, scheme: str, netloc: Optional[str]) -> FileSystem:
         from pyarrow.fs import HadoopFileSystem
 
         hdfs_kwargs: Dict[str, Any] = {}
-        replication = self.properties.get(HDFS_REPLICATION, "0")
+        replication = self.properties.get(HDFS_REPLICATION, "3")
         hdfs_kwargs["replication"] = int(replication)
         if netloc:
             return HadoopFileSystem.from_uri(f"{scheme}://{netloc}/?replication={replication}")