On my machine it's not G29 that is used to measure the bed probe errors, it's a macro file. The log extract below reads from bottom to top:
21:40:03 M98 P/macros/Check bed
Bed probe heights: -0.025 -0.017 0.020 -0.002 0.027 -0.012 0.013 0.036 0.005 -0.023 -0.021 0.021 -0.009, mean 0.001, deviation from mean 0.020
21:39:13 M98 P/macros/Check bed
Bed probe heights: -0.019 -0.010 0.020 0.011 0.021 -0.012 0.013 0.042 0.012 -0.005 -0.014 0.033 -0.009, mean 0.006, deviation from mean 0.019
21:38:36 G32
Calibrated 6 factors using 13 points, deviation before 0.202 after 0.016
21:37:57 Connection established!
Looks like about 0.018mm in the worst case (i.e. 3 motor microsteps), but more usually 0.006 (1 microstep).
Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod
Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].