Linux (at least) builds fail on tests where you don't actually solve anything

[reverendbedford]

I get either nice errors or segfaults from pardiso on Linux when running tests where all the DOFs are constrained. An example is example_problem_threads/test_17. The failures occur when you call pardiso over an "empty" system of equations, as there is nothing to solve for in this example.

An easy fix is to guard against this condition and just return from the solve manager if the system is empty.