This caused an infinite loop for "surfacing" / dropcutter operations.

Previously the parameters for grid/spiral path patterns were invisible
for new processes. It was necessary to switching the path pattern type
or the path strategy to work around this issue.
Now another signal is emitted to unhide the path pattern paramters.

* Windows: use "urllib.url2pathname" instead of manual decoding
* non-Windows: use "urllib.url2pathname" instead of doing nothing

Filenames including whitespace or special characters can now be used
with the RecentChooserMenu ("recently used model files").

thanks to Valerio Bellizzomi for reporting this issue!

thanks to Valerio Bellizzomi for reporting this issue

Previously the "TinySidewaysMoveFilter" was used to detect small gaps within
a toolpath that can be removed safely. But the above filter only worked under
the specific conditions that the z levels of both adjacent points are the same.
The new implementation adds another connection line to the motion grid (before
collision tests). Thus now all kind of connections (independent of their length
and position) are possible.

* hide "description" column of processes (not in use, yet)
* trigger control update for name changes of model/tool/process/bounds

Just one varible name was used improperly. This broke the "waterline" strategy.

This is a useful feature for debugging path generator exceptions. Otherwise
the exception's backtraces are concealed by multiprocessing function calls.

Currently task settings are not implemented.

Currently it is not possible to switch between different postprocessors.
The necessary code will be added later.

* move tiny side move filter _before_ the safety heigth filter
* simplify the code by using pnear instead of manual point comparisons

SafetyHeightFilter:
* skip move to safety height if only the height of a position changed
* simplified code

StepWidth:
* just skip small steps; don't change output formatting
* the output conversion will be done later in a more appropriate place

* before: just after the whole run - this was not enough for separate lines

* otherwise the events are not triggered after manual input (e.g. for safety height)

* this function just applies the usual "floats are within range" comparison

* this is based on a dirty copy'n'paste code piece that needs to be cleaned

* the closing move to safety height was simply forgotten
* additionally a (rarely triggered) branch was broken before

* fixed simulation speed (minutes/seconds issue)
* dynamically change speed factor step width