OpenRSP test ggf_lda: Harmonize thresholds between the two tests

There are two tests for ggf_lda with two different calculation rules. Some work was done to chg them because there was an error for these tests which now appeared to still persist on the dashboard. This error was due to one value sometimes coming slightly above the write threshold, thereby causing a mismatch in the number of elements written to rsp_tensor between the test result and the reference.

I have changed the write threshold for the (k,n) = (0,2) test to 1e-7 which was the same threshold as was already used for the (k,n) = (1,1) test. At the same time I changed the rsp eqn convergence threshold to 1e-8 so it is the same between these two tests (was 1e-7). Possibly it was intended to be equal between both tests but by a small mistake was done opposite in one.

I haven't tested this but the change is miniscule and there will be a pipeline for this MR. The real test however lies in the behavior on the dashboard (which is where this behavior caused the tests to fail) and we will see that once this is merged and run there.

Nominating @foeroyingur to review since you were the reviewer the prev time this test was changed by Simen.