Fix of #5007 - "Reload from disk" causes objects converted to inches to revert to mm
1) Storing and reloading the "source_in_inches" source flag from AMF and 3MF
2) When converting objects with mixed "inches" volumes, do the right thing
and do not convert those that do not need conversion.
This is a fix of a bug, which was in Slic3r forever, where raw slices
were not cached, but recalculated from classified regions, where
merging the regions did not produce the original contour reliably.
Fixes [2.3.0-beta2] Odd bad slicing related to infill (?) percentage #5407
Replaced all "unsigned long long" types with uint64_t.
These "new" types better communicate their meaning and they are
guaranteed to behave the same on all platforms & compilers.
(linear tapering of extrusion rate for the 1st spiral vase layer)
Linear tapering could only be implemented for relative extruder distances.
For absolute extruder distances, all the G1 Exxx in all spiral vase layers
would have to be adjusted, which is much more complex and risky to do now
before release.
above the solid infill. Should fix
first layer of vase (after bottom layers) is bulging (0.6nozzle, 0.4 layerhight) #3712
Adjust flow of the "second" layer in spiral vase mode #2795
(non-spiral vase layers) were being closed.
Fixes Bad slicing in vase mode (unexpected bridge and solid infill layers) #3326
Fixes Model with holes in the base does not slice properly in "Vase Mode" #5359
The thread local storage variables for the random number generator
were replaced by an external context variable.
Thread local storage is a scarce resource.
1) Decreased "gimbal lock" epsilon 10x
2) Got rid of unnecessary back and forth conversions which led to numerical inacurracies
This should fix issues #5172, #5011 and #5398
Bounding boxes of polygons could overlap. Ask the AABB tree for all possible candidates.
Might be faster than searching for the closest triangle, that requires traversing the whole depth of the tree every time.
