Here's two fixes to tests to get them to work properly on Frontier
HLLD Double Star State Test
This test was testing uninitialized data and I just didn't notice until now. Now the data is properly initialized.
MHD Blast Wave System Test
This test had differences compared to and identical run on C-3PO at the ~1% level in a small handful of cells. I'm pretty sure that this is just an issue floating point differences propagating so I added an option to test with the L2 norm to data based system tests instead of having to do a cell-by-cell comparison. This should be less sensitive to a handful of outliers.
Here's two fixes to tests to get them to work properly on Frontier
HLLD Double Star State Test
This test was testing uninitialized data and I just didn't notice until now. Now the data is properly initialized.
MHD Blast Wave System Test
This test had differences compared to and identical run on C-3PO at the ~1% level in a small handful of cells. I'm pretty sure that this is just an issue floating point differences propagating so I added an option to test with the L2 norm to data based system tests instead of having to do a cell-by-cell comparison. This should be less sensitive to a handful of outliers.