gs2_optimisation test seems broken

[Ryusuke Numata]

When I run gs2_optimisation test, it fails with 6 mpi processes. I think this happens on any hosts. At least I've confirmed this happens on my mac desktop, linux cluster (rokko).

The problems is invalid array access errors occuring at transform2_5d_accel, eg

At line 1119 of file gs2_transforms.f90 Fortran runtime error: Index '24' of dimension 3 of array 'g' above upper bound of 23

At some point of the test, transform2_5d_accel is called without calling init_accel_transform_layout.

This happens probably because the accelerated flag is wrongly set to true. I'm not sure about the behavior of gs2_optimisation, but it seems performing several different layout runs sequentially without appropriately initialize gs2_transform. So accelerated flags seems not initialized, and is carried over.

[Joseph Parker]

Thanks - I've seen this sporadically on the test server, but not on my local machine. I'll look at it when the Jenkins test server becomes a bit more stable

[David Dickinson]

This test has now been deleted. A workaround seemed to be to prevent the number of processors from being varied as a part of the test.