[quant][graphmode] Add new tensorlist observer for LSTM (#35893)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/35893
LSTM operator inputs have tensor list for activations and weights.
In graph mode we need a new observer to work with tensor list
Test Plan:
python test/quantization/test_quantization.py ObserverTest
Imported from OSS
Differential Revision: D20830389
fbshipit-source-id: 4790f8932ae3d38446c1d942a2b3780aa91e3022