fix: Automatically send truncated long ints to cuda at shape analysis time

[gs-olive]

Description

• Augment aten::to operator insertion at shape analysis time to insert the target device
• Add functionality to PartitioningInfo struct to store device information and produce cuda device string
• Make cuda device string function in LowerInfo and PartitioningInfo const to avoid altering struct fields

Uses schema

Tensor.to(device : Device,
          dtype : int,
          non_blocking : bool=False,
          copy : bool=False,
          memory_format : Optional[int]) -> Tensor

Instead of

Tensor.to(dtype : int,
          non_blocking : bool=False,
          copy : bool=False,
          memory_format : Optional[int]) -> Tensor

This switch was made to ensure the device for truncated objects is GPU, regardless of their origin, to avoid adding another lowering pass for this case. Since an aten::to operation is already being inserted, use the opportunity to use correct tensor device (GPU).

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• [ x ] My code follows the style guidelines of this project (You can use the linters)
• [ x ] I have performed a self-review of my own code
• [ x ] I have commented my code, particularly in hard-to-understand areas and hacks
• [ x ] I have made corresponding changes to the documentation
• [ x ] I have added tests to verify my fix or my feature
  • Tested on a few models, sample scripts + existing fallback test suite
• [ x ] New and existing unit tests pass locally with my changes
• [ x ] I have added the relevant labels to my PR in so that relevant reviewers are notified

[bowang007]

LGTM