fix: use pyright-specific ignore for DataFrame columns parameter

Revert to using columns parameter with correct pyright ignore syntax
(reportArgumentType) instead of attempting to set columns property
after creation, which fails for empty DataFrames.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: goloroden

eventsourcingdb/pandas.py

@@ -40,23 +40,23 @@ async def events_to_dataframe(events: AsyncGenerator[Event, None]) -> pd.DataFra
         }
         event_list.append(event_dict)
 
-    df = pd.DataFrame(event_list)
-
     if len(event_list) == 0:
-        df.columns = pd.Index([
-            "event_id",
-            "time",
-            "source",
-            "subject",
-            "type",
-            "data",
-            "spec_version",
-            "data_content_type",
-            "predecessor_hash",
-            "hash",
-            "trace_parent",
-            "trace_state",
-            "signature",
-        ])
-
-    return df
+        return pd.DataFrame(
+            columns=[  # pyright: ignore[reportArgumentType]
+                "event_id",
+                "time",
+                "source",
+                "subject",
+                "type",
+                "data",
+                "spec_version",
+                "data_content_type",
+                "predecessor_hash",
+                "hash",
+                "trace_parent",
+                "trace_state",
+                "signature",
+            ]
+        )
+
+    return pd.DataFrame(event_list)