update optimizer type hint

Author: awaelchli

pytorch_lightning/plugins/precision/native_amp.py

@@ -108,7 +108,7 @@ def pre_optimizer_step(
         return False
 
     def post_backward(
-        self, model: "pl.LightningModule", closure_loss: torch.Tensor, optimizer: torch.optim.Optimizer
+        self, model: "pl.LightningModule", closure_loss: torch.Tensor, optimizer: Optimizer
     ) -> torch.Tensor:
         ret_val = super().post_backward(model, closure_loss, optimizer)
         # unscale here to have it inside the closure before the grad tracking and clipping