Add tensorflow 2 conversion to tensorrt

[cloud-rocket]

https://github.com/NVIDIA/TensorRT/issues/513

References:

• https://gilberttanner.com/blog/run-tensorflow-on-the-jetson-nano

[cloud-rocket]

via onnx:

apt-get install libprotobuf-dev protobuf-compiler
pip3 install -U tf2onnx

inputs / outpus - Inputs = [['img_in']], Outputs = [['n_outputs0/BiasAdd', 'n_outputs1/BiasAdd']]

#python3 ~/projects/donkeycar/scripts/freeze_model.py --model=~/mycar/models/linear.h5 --output=~/mycar/models/linear.pb

#python3 -m tf2onnx.convert --input ./models/linear.pb --inputs img_in:0 --outputs n_outputs0/BiasAdd:0,n_outputs1/BiasAdd:0   --output ./models/linear.onnx

 /usr/src/tensorrt/bin/trtexec --onnx=./models/linear.onnx