Support QNN (Qualcomm AI Engine Direct)

[anyj0527]

May make tensor_filter::qnn? or support it as other filter's delegate (tflite / onnxruntime / executorch)
QNN: The Qualcomm® AI Engine Direct SDK provides lower-level, unified APIs for AI development.

REF: https://developer.qualcomm.com/software/qualcomm-ai-engine-direct-sdk

[taos-ci]

cibot: Thank you for posting issue #4296. The person in charge will reply soon.

[anyj0527]

tflite's QNN delegate: in some private place
onnxruntime's QNN Execution Provider: https://onnxruntime.ai/docs/execution-providers/QNN-ExecutionProvider.html
executorch's QNN backend: https://github.com/pytorch/executorch/blob/main/backends/qualcomm/setup.md

[myungjoo]

If it's going to be a delegation of a subplugin (tflite, onnxruntime, ...), as long as we open up an interface for the runtime/subplugin, it should be enough.

Anyway, are you going to use accelerator property for configuring delegation ? (then, we need additional comments on the property description and the subplugin documentation (nnstreamer-chk).)