Simple test passes already

Avoid explicit doubles.

The check fails in recent CI, but I think it should be correct
anyway. Also, I dont think that error bounds below single precision
accuracy are good for numdiffs.

No substantial changes there.

They are not used in a senseful way anyway and loopy
stumbled over them.

This is currently only used for level 4 instrumentation,
as these start and stop markers were messing heavily with
dependencies in the sum factorization code. This way, the necessary but
ugly work of adding instrumentation is separated much more
nicely.

I was tired of the amount of ifs that changed behaviour
depending on stage 1 or 3. This is a cleaner approach.

This is a temporary fix for CI. After updating to
loopy 2018.1, this should be properly fixed.

It is not necessary anymore

It is applied to the function call...