bugs in algs::MethodOfLinesIntegrator

[humhang]

Hello, in algs/MethodOfLineIntegrator.C:Line252, the ghost cells at the level ln are filled by the patch data in the coarser level ln-1, but the patch date at that coarser level has already been updated to the next stages.

that's to say, you are trying to fill the ghost cell with data that has a different time stamp.

Also, the data should be coarsen from the finest level to the coarsest level, that's to say, Line 276 should be called from ln = nlevel-1 to 1. However, it is called from 1 to nlevel-1.

[nselliott]

Your report is correct here, so we will be fixing this.