[NDArray] Map NDArray ops to GPUs

[tkarna]

Adds machinery to support GPU pipeline in Sharded Array for Python.

• InsertGPUAllocsPass: add host-shared option, defaults to true.
• Mark ndarray.copy ops with a special region environment ("gpu_copy_op") if either operand has a GPU environment.
• Adds InsertGPUCopyPass to convert memref.copy ops within GPU env with gpu.memcpy ops.
• Adds TileLoops pass to tile loops ops with tileUsingSCF. Apply twice to map to blocks and threads.
• Adds RegionParallelLoopToGpu to convert parallel loops to GPU only if within a gpu env region.

---

Please review these guidelines to help with the review process:

• Have you provided a meaningful PR description?
• Have you added a test, a reproducer, or a reference to an issue with a reproducer?
• Have you tested your changes locally for CPU and GPU devices?
• Have you made sure that new changes do not introduce compiler warnings?
• If this PR is a work in progress, are you filing the PR as a draft?
• Have you organized your commits logically and ensured each can be built by itself?