1st performance pass

[timzuntar]

Identify suboptimal resource usage and try to improve it

[timzuntar]

Vectorizing what can be vectorized ist the first bullet point on the agenda.

[timzuntar]

It seems like closing the issue by post-commit comment did not work, so here it is.

While optimization is still not stellar and better ways of doing calculations can certainly be found, the runtime of the test example (generating a single image generated by standard microscopy and comparing it with a STED-enabled one) has for the combination of parameters used been reduced from 48 seconds to 65 milliseconds, an improvement by a factor of some 700. Good enough for now.