Fill CUDA opset gap for Softplus and Softsign to opset 22

[elwhyjay]

Closes part of #27729.

The CUDA EP registered Softplus and Softsign only for opset 1 (HFD). Opset 22 extended the type constraint to include BFloat16, but the CUDA EP was never updated. The BFloat16 compute kernels were already instantiated via UNARY_ACTIVATION_OP_HFD_WITH_BF16; only the EP registration was missing.

Changes:

• Cap opset-1 kernels as versioned [1, 21] (Softplus, Softsign)
• Register opset-22 kernels with BFloat16 support (HFDX)
• Add opset-22 tests: float (CPU+CUDA) and BFloat16 (CUDA, sm≥530)
• Update docs/OperatorKernels.md

[Copilot]

Pull request overview

This PR updates ONNX Runtime’s CUDA Execution Provider to correctly register Softplus and Softsign kernels through opset 22+, matching the ONNX schema change that extends type constraints to include BFloat16 (while keeping existing kernels valid for opsets 1–21). It also adds regression tests and updates the generated operator-kernel documentation accordingly.

Changes:

• Version-cap CUDA Softplus/Softsign kernel registrations to [1, 21] and add 22+ registrations including BFloat16.
• Add opset-22 unit tests for Softplus/Softsign (float) and CUDA-only BF16 coverage.
• Update generated docs/OperatorKernels.md entries to reflect the new registration ranges/types.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
onnxruntime/test/providers/cpu/activation/activation_op_test.cc	Adds opset-22 tests for Softplus/Softsign including CUDA BF16 cases.
onnxruntime/core/providers/cuda/cuda_execution_provider.cc	Updates CUDA kernel registration ranges and adds opset-22 typed registrations with BF16.
onnxruntime/core/providers/cuda/activation/activations.cc	Adjusts CUDA activation kernel instantiations to split opset 1–21 vs 22+ (BF16).
docs/OperatorKernels.md	Updates the generated CUDA EP kernel table entries for Softplus/Softsign opset ranges and supported types.