Initial pr to support dynamic shape detection

[JackCaoG]

The implementation in this pr is not ideal because it can only handle a single graph for a compiled program. This is usually too strict for the real world program because in training we usually at least have 2 graphs since in the step0 the optimizer state is not initialized.

I think what I need to do next is to provide a max_dynamic_graph_allowed arguments and use a TRIE to keep tracked of all the IR being generated and when they diverage.

[JackCaoG]

rioght now if I used this feature in https://github.com/pytorch/xla/blob/master/examples/train_decoder_only_base.py it will error out because second step graph is different from the first step one(most likely due to the optimizer state).