A cleaner and more scalable implementation of symbolic tracing (#11763)

Cleaner and more scalable implementation of symbolic tracing with torch.fx, and provides support for new architectures:
- ALBERT
- DistilBERT
- MobileBERT
- MegatronBERT
- GPT2
- GPT Neo

Co-authored-by: Michael Benayoun <michael@huggingface.co>

tests/test_modeling_megatron_bert.py

@@ -1,4 +1,3 @@
-# coding=utf-8
 # Copyright 2021 The HuggingFace Inc. team. All rights reserved.
 # Copyright 2021 NVIDIA Corporation. All rights reserved.
 #
@@ -282,6 +281,7 @@ class MegatronBertModelTest(ModelTesterMixin, unittest.TestCase):
         if is_torch_available()
         else ()
     )
+    fx_ready_model_classes = all_model_classes
 
     # test_resize_embeddings = False
     test_head_masking = False