Describe the bug
CUDAExecutionProvider does not handle Clip on float16. It gets resorted to CPUExecutionProvider.

Urgency
None. But prefer it to be fixed.

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Ubuntu 20.04
• ONNX Runtime installed from (source or binary): pip
• ONNX Runtime version: 1.8.0
• Python version: 3.8.11
• Visual Studio version (if applicable): N/A
• GCC/Compiler version (if compiling from source): N/A
• CUDA/cuDNN version: 11.2
• GPU model and memory: RTX 2060 Super

To Reproduce
Unable to provide a model right now. You can create a graph with a clip node with float16 input, min, and max value.

Expected behavior
Operation executed on CUDA