What is the difference between VISION_POSITION_ESTIMATE and GLOBAL_VISION_POSITION_ESTIMATE messages ? And local vs global x,y,z coordinates?
As usual, wiki is your friend 
https://mavlink.io/en/messages/common.html#GLOBAL_VISION_POSITION_ESTIMATE
We dont consume GLOBAL_VISION_POSITION_ESTIMATE messages within Ardupilot as we set local reference to EKF Home position using mavlink message or by right click on the map within Mission Planner
So, how about VISION_POSITION_ESTIMATE local x,y,z coordinates. What does it mean “local” x,y coordinates ? Does it mean x,y distance from home position to the drone?
You set the EKF home position in order to get the system started
Once again WIKI is your friend:
https://ardupilot.org/dev/docs/ros-vio-tracking-camera.html