onnxruntime

Enable static memory planning for pipeline. (#4204)

* Enable static memory planning for pipeline.
1. We fix a bug when resolving symbolic shape for scalars.
2. We pass the original inputs to all pipeline stages so that
   the symbolic shapes can be resolved.

* Further Improvements
1. Address comments.
2. Further reduce activation size by ~50% when pipeline is on.
   This is done by removing all but one gradient tensor from the last
   RecordEvent in the backward pass.

* Address a comment

* Fix Windows build