How to fix APM's 1m maximum accuracy?

The issue of Mavlink/APM’s use of floats instead of scaled integers (link1) has been brought up before here but never resolved. There is already a framework to fix this by using COMMAND_INT and/or MISSION_ITEM_INT. (see links below) There also seems to be some progress on the 3DR Solo that might also be applicable to APM 3.4. Solo COMMAND_INT / Solo MISSION_ITEM_INT

MISSION_ITEM_INT receiving was added to MissionPlanner a few days ago.

Any help or advice toward an accurate APM would be greatly appreciated.!topic/drones-discuss/ZVIHKqsPipQ!topic/px4users/eEZNedi7lBg

Thanks to Michael Oborne Mission Planner now has support for MISSION_ITEM_INT and Ardupilot support is pending a pull request here.

Would the Emlid team mind helping get the bugs out and get it merged so we can finally have cm accurate way points?