Support NumPy 2.0 by replacing np.NaN with np.nan

Fixes: #19

Ref: https://numpy.org/devdocs/numpy_2_0_migration_guide.html#main-namespace

tbparse/summary_reader.py

@@ -849,7 +849,7 @@ def _get_default_cols(self, tag_to_events: Dict[str, ScalarEvent],
         length = 0
         for events in tag_to_events.values():
             length += len(events)
-        cols: Dict[str, Any] = defaultdict(lambda: [np.NaN] * length)
+        cols: Dict[str, Any] = defaultdict(lambda: [np.nan] * length)
         if 'dir_name' in self._extra_columns:
             cols['dir_name'] = [''] * length
         if 'file_name' in self._extra_columns: