[ Cuda] ConvTranspose-22 (#27710)
## Description
This PR extends the CUDA `ConvTranspose` operator registration to
support ONNX opset 22. The CUDA implementation already shares the same
attribute and output-shape handling used by earlier supported opsets, so
this change primarily exposes the existing kernel path for opset 22 and
adds regression coverage to keep that path working.
## Summary of Changes
### CUDA Kernel Registration
| File | Change |
|------|--------|
| `onnxruntime/core/providers/cuda/nn/conv_transpose.cc` | Split
`ConvTranspose` kernel registration into `1-10`, `11-21`, and `22` so
the CUDA kernel can be selected for opset 22. |
| `onnxruntime/core/providers/cuda/cuda_execution_provider.cc` | Added
matching provider-side kernel declarations and registry entries for
`ConvTranspose(22)` on CUDA. |
| `onnxruntime/core/providers/cuda/cuda_nhwc_kernels.cc` | Added
matching NHWC CUDA declarations and registry entries for
`ConvTranspose(22)`. |
### Test Coverage
| File | Change |
|------|--------|
| `onnxruntime/test/providers/cpu/nn/conv_transpose_op_test.cc` | Added
a CUDA-only opset 22 regression test that validates `ConvTranspose` with
`output_shape`. |
| `onnxruntime/test/providers/cuda/nhwc/conv_transpose_test.cc` |
Updated existing NHWC comparison coverage to instantiate `ConvTranspose`
at opset 22. |
## Testing
- Built the touched translation units successfully with targeted `ninja`
object builds:
- `onnxruntime/core/providers/cuda/cuda_execution_provider.cc`
- `onnxruntime/core/providers/cuda/cuda_nhwc_kernels.cc`
- `onnxruntime/core/providers/cuda/nn/conv_transpose.cc`
- `onnxruntime/test/providers/cpu/nn/conv_transpose_op_test.cc`
- `onnxruntime/test/providers/cuda/nhwc/conv_transpose_test.cc`
- Ran `git diff --check` to confirm the patch is formatting-clean.
- A full `onnxruntime_test_all` / gtest runtime pass was not completed
locally because the build tree regenerated and expanded into a broad
rebuild; runtime verification of the new CUDA and NHWC tests should
still be run in CI or a focused local test build.
## Motivation and Context
Related issue: https://github.com/microsoft/onnxruntime/issues/26393
The ONNX `ConvTranspose-22` schema keeps the same core padding and
output-shape semantics used by the earlier CUDA-supported opsets, but
CUDA registration stopped at opset 11. That meant models using
`ConvTranspose` at opset 22 could miss CUDA kernel assignment even
though the implementation path was already compatible.
This PR closes that gap by updating kernel registration and test
coverage without changing the underlying CUDA compute logic.
## Checklist
- [x] Tests added/updated
- [ ] Documentation updated (if applicable)
- [x] No breaking changes (or documented in description)
- [ ] CI passes
