test timing reference.
The developers have a feeling for the individual test run times, the user does not. Sometimes I have the impression the tests are very slow for some reason, maybe because something with my compilation is not optimal.
Small suggestion:
In the test output, provide the timing for a 'standard desktop computer of ours, or ever for the usual machines' behind the runtime of the test. If the test runtime is much larger then this 'benchmark' on a 'performance' resource, something might be fishy, or one has compiled with gfortran instead of intel. Also it might help developers to track the evolution of the tests duration a bit better, because the timings are in the git.
I am not sure if this is a good idea through.