To address KGV comments the approach has been changed to only measure BRepMesh part as
follows:
- for each shape a deflection has been computed from the bounding box as
  d=0.004 * MAX(dX, dY, dZ) and rounded, to match OCC algorithm (GetDeflection() in StdPrs_ShadedShape.cxx)
- bounding box received with the DRAW command 'bounding s' after 'tclean s'
- you can argue that incmesh mesh use another angular deviation than visualization. I believe we can
  disregard that given that the values are equal between the versions.
  Please advise if there is any better way to isolate the BRepMesh in DRAW.

The list of DRAW commands:
tclean s; chrono c reset; chrono c start; incmesh s <val>; chrono c stop; chrono c show
Minimum of 5 runs chosen

-------------------------------------------------------------------
shaver:
compound of subshapes from http://www.opencascade.org/showroom/shapegallery/gal3/

tclean s; chrono c reset; chrono c start; incmesh s 0.68; chrono c stop; chrono c show

6.3.1:
Elapsed time:  0 Hour(s) 0 Minute(s) 0.87 Second(s)
CPU user   time: 0.8736056 seconds
CPU kernel time: 0 seconds

6.5.4:
Elapsed time: 0 Hours 0 Minutes 1.42906735078 Seconds
CPU user time: 1.4196091 seconds
CPU system time: 0 seconds
-------------------------------------------------------------------

Cylinder_Head:
http://www.opencascade.org/showroom/shapegallery/gal4/
http://www.opencascade.org/ex/att/15_cylinder_head.brep.gz

tclean s; chrono c reset; chrono c start; incmesh s 0.74; chrono c stop; chrono c show

6.3.1:
Elapsed time:  0 Hour(s) 0 Minute(s) 0.746 Second(s)
CPU user   time: 0.7488048 seconds
CPU kernel time: 0 seconds

6.5.4:
Elapsed time: 0 Hours 0 Minutes 1.47951063298 Seconds
CPU user time: 1.4508093 seconds
CPU system time: 0 seconds
-------------------------------------------------------------------

F1
http://www.opencascade.org/showroom/shapegallery/gal5/
Disregarded due to substantial mesh quality (6.5.4 is far better than 6.3.1)