Generally, calibration issues with the IR probe come down to two issues:
Effector tilt- You have the probe offset from the nozzle by some distance, however due to the geometric shortcomings of your machine, the effector may be tilting very slightly at different x y positions (ie rods not space properly at both ends etc), because the probe is offset and the effector is tilting, it may be closer to the bed at one xy position than another.
IR interference - reflective surfaces etc. Sounds like this isn't the issue. I'm sure the experienced gang here will help you sort out the gcode/config confusion you're having, but if that doesn't work out, use a spirit/target level on the effector to see if you're dealing with effector tilt. You can use the H parameter in your bed.g to attempt to calibrate out effector tilt, but I've had mixed results doing so personally, best to address why your effector is tilting if that turns out to be what's happening.
