FaceONNX models import issues tracker

[SimonBrandner]

This is a tracker issue for problems with importing models from FaceONNX. Here are some useful links:

• FaceONNX repo
• FaceONNX models folder on Google Drive
• My test repo

ToDo list:

• eye_blink_classifier
• recognizer
• detector: almost done (see Trouble importing FaceONNX detector model: Only tensor indices is valid #1915)
• depth_classifier: ONNX import issue: index out of bounds in pad_config() due to node.input only having len of 2 #2111, ONNX import issue: ceil_mode is not supported in avg_pool1d_config() #2132
• landmarks_detector: ONNX import issue: Reshape: shape tensor must be present for Node #2115
• yolov5s-detector: ONNX import issue: Reshape: shape tensor must be present for Node #2115