extend test coverage in time_advance-tests

[bmcdanie]

currently, we test only that the wavelet space solution for each problem at a given timestep matches the MATLAB solution for that problem/timestep. we could fairly simply extend these tests to also test that the difference between wavelet space solution and analytic solution (for problems with analytic solution) also matches. this will ensure we have consistent analytic solutions generated at every timestep.

[hbrunie]

Are these tests extended to double and single precision ? (are we testing ASGARD_DOUBLE_PRECISION OFF)?

[bmcdanie]

yes - that macro is only used to control precision in main, but time advance tests are templated to run in single and double precision.