Rayleigh_Ray2, PnPn-2 failure

[nek5000]

Rayleigh_Ray2, PnPn-2 failure

[stgeke]

I think the delta of 1e-5 is too tight for regression testing. Even small variations (e.g. disabling residual projection, different machine) produce a slight different result in order of 1e-5. I get same test value of 6.621518E-03 using the current master and the latest svn version.

I recommend to set delta to 1e-4.

[stgeke]

I can reproduce the test value of 0.006728787with r1073 on my machine. The latest SVN version (r1142) produces 6.621518E-03.

Note, this cases uses estimated varying residual for the pressure and Helmholtz solve. If I use fixed residual tolerances I get the same test value for both revisions.

Instead of relaxing delta we could also set fixed residual tolerances.

[RonRahaman]

Fixed in PR #66