Add eIQ Neutron NPU custom operator into TFLM kernels

[nxpxf]

I'm from NXP and working on NXP eIQ Neutron NPU support in TFLM. We had supported Neutron NPU in TFLM through downstream approach (https://github.com/nxp-mcuxpresso/mcuxsdk-middleware-tensorflow/tree/main/tensorflow/lite/micro/kernels/neutron).

Now we want to support neutron kernel in TFLM community, we plan to follow similar approach as our current implementation, but keep dependent neutron-driver and neutron-firmware library in out NXP github repo (https://github.com/nxp-mcuxpresso/mcuxsdk-middleware-eiq/tree/main/neutron).

May I know your feedback whether this approach is acceptable by TFLM community? If yes, what's the process to submit patch to TFLM community for adding this new NPU kernel?

Appreciate for your feedback and guidence on this.