Add test for the DD interval scalar solver

I reused a part of the old 'regular' solver tests and added support for applying different solution intervals per direction. The test assigns random solutions to every different solution-interval, and the visibilities of the intervals are modified accordingly.

The solver seems to work :). Somewhat to my surprise I did not have to increase the tolerance; the solver returns very accurate results even though the degrees of freedom is increased by quite a bit. I'm using 1,2,5 solutions for direction 0,1,2 respectively, so it uses now 8 solutions/chblok instead of 3.

I was a bit suspicious by how accurate the results are, so to test the test I tried lowering the number of maximum iterations. This indeed shows the test are correct, and e.g. using 50 instead of 100 iterations shows a few solutions that haven't been converged yet, but which are clearly going into the right direction. So, all looks good.