Fix validation on restart

Author: lantiga

src/lightning/pytorch/loops/training_epoch_loop.py

@@ -208,6 +208,13 @@ def advance(self, data_fetcher: _DataFetcher) -> None:
             StopIteration: When the epoch is canceled by the user returning -1
 
         """
+        if (
+            self.restarting
+            and self._should_check_val_fx(data_fetcher)
+            and self.val_loop.batch_progress.total.started > self.val_loop.batch_progress.total.completed
+        ):
+            # skip training and run validation in `on_advance_end`
+            return
         # we are going to train first so the val loop does not need to restart
         self.val_loop.restarting = False