perf: skip fill value scan and copy for variables without fill values

Only copy and scan for fill values when the variable actually has a
_FillValue attribute. Most variables default to NaN, making the
comparison a no-op that still scans the entire array. Variables
without fill values now return a zero-copy view from reshape.

Author: berkgeveci

src/e3sm_quickview/plugins/eam_reader.py

@@ -423,8 +423,10 @@ def _load_variable(self, vardata, varmeta):
                     slice_tuple.append(self._slices.get(dim, 0))
 
             # Get data with proper slicing
-            data = vardata[varmeta.name][tuple(slice_tuple)].data.reshape(-1).copy()
-            data[data == varmeta.fillval] = np.nan
+            data = vardata[varmeta.name][tuple(slice_tuple)].data.reshape(-1)
+            if not np.isnan(varmeta.fillval):
+                data = data.copy()
+                data[data == varmeta.fillval] = np.nan
             return data
         except Exception as e:
             print_error(f"Error loading variable {varmeta.name}: {e}")