AC_Avoid: add support for MNT EYE depth cameras

ArduPilot already supports object avoidance using the Intel RealSense 435 and 455 depth cameras. We should also add support for the [MYNT EYE depth cameras](https://www.mynteye.com/pages/products).

This project involves:

- Writing a python script ([similar to this script for the T435](https://github.com/thien94/vision_to_mavros/blob/master/scripts/d4xx_to_mavlink.py)) to pull the data from the depth camera and package them into OBSTACLE_DISTANCE and/or OBSTACLE_DISTANCE_3D mavlink messages which will then be consumed by AP's AP_Proximity library
- Creating an APSync image for at least one companion computer (RPI4?) that can run the above script

Start by reading the linked vision_to_mavros scripts/d4xx_to_mavlink.py and the AP_Proximity library entry point. Determine how MYNT EYE depth data can be packaged as OBSTACLE_DISTANCE or OBSTACLE_DISTANCE_3D messages, then investigate building an APSync image for an RPI4 companion computer. Done means the camera data is consumed for object avoidance and the image runs the script.

Written by the indexing model from the issue text.