Sounds pretty good to me, the standard deviation of the set of bed probing points you have is between 1 and 2 hundredths of a millimetre. I think that will probably do the trick. I am getting down to 0.02 usually and my printer is fairly "no compromise". How does a bed level test spiral across the entire bed look? Are your calibration objects dimensionally accurate?
My other delta, a much less rigid machine, manages to using Rich Cattell's marlin, derive a solution to 0.05mm after about 30 minutes of trying.
I think the only thing which would enhance this accuracy would be to then grid level, to compensate for any unevenness in the bed. But I think this feature is planned for the 1.17 firmware release.