Update testRandom

I updated the testRandom example by replacing the criteria numbers we used by the standard errors expected (times 5 for 5 sigma), this should fix this test from failing. I tested it and so far it doesn't seem to fail anymore.

This should fix #607 (closed) completely

Have you had a chance to try with the compiler setting that usually breaks on cosma?

Yes, it doesn't seem to crash anymore in this case. I tried around 20-30 times if it would break and it doesn't break anymore as before.

Great. Thanks for taking care of it.

merged

mentioned in commit 58dfc0f9