scan.peak and scan.com are rounded too much using axis.tolerance
For the ID11 monochromator, with a peak at 2.6034 degrees then Bliss goes to 2.6 degrees and misses the peak altogether. The axis.tolerance of 0.01 was set to match the measured backlash and the axis uses an encoder. The problem is in bliss.scanning.scan.Scan methods _peak and _com.
https://bliss.gitlab-pages.esrf.fr/bliss/master/motion_axis.html#tolerance
Axis tolerance is the acceptable discrepancy between controller position and last known axis dial position when starting a move.
Related to #1487 (closed) and #1592 (closed)