The only function using the last_retraction_config was the function which owuld convert a normal retraction into a nozzle switch retraction, instead of first performing the normal retraction and then the nozzle switch retraction. That functionality had to be removed anyway, because performing the nozzle switch retraction on the print can leave more scarring.
The test currently fails due to a floating point rounding error. I'll give it some lenience, especially since both the ground truth and the supposed output are estimates.
It will set up a very basic g-code planner and let it estimate the time and material it needs. Currently it only tests with empty g-code, but I mean to add a test with retraction once this works. Currently the setup gives a segfault.
Contributes to issue CURA-415.
Tim Kuipers