Slope limiting to corner of cell, not fine cell next to corner of cell

[sw561]

Summary

Proposed change to illustrate the issue being described in #4382
Slope limiting is based on value in the fine cell (and thus dependent on refinement ratio) instead of the corner of the cell.

Additional background

See issue #4382

Checklist

The proposed changes:

• fix a bug or incorrect behavior in AMReX
• add new capabilities to AMReX
• changes answers in the test suite to more than roundoff level
• are likely to significantly affect the results of downstream AMReX users
• include documentation in the code and/or rst files, if appropriate

[WeiqunZhang]

I am reluctant to accept the changes. It might affect a lot of codes with unknown consequences. Maybe we could add a new option for this instead of replacing the current behavior.

[asalmgren]

Also -- imagine a bilinear function -- the slope should be limited in that case by the fine corner ghost cell, not the corner of the cell