[[Imported from SVN: r6792]]

[[Imported from SVN: r6779]]

After all, we are thinking 'Steklov-Poincare' here, which means
we are solving an equation on the interface.  The old code
taking into account the subdomain solutions wasn't really satisfying:
in particular, it was not symmetric wrt the subdomains.

[[Imported from SVN: r6762]]

[[Imported from SVN: r6761]]

[[Imported from SVN: r6760]]

[[Imported from SVN: r6759]]

[[Imported from SVN: r6758]]

[[Imported from SVN: r6757]]

[[Imported from SVN: r6756]]

[[Imported from SVN: r6755]]

[[Imported from SVN: r6754]]

[[Imported from SVN: r6753]]

[[Imported from SVN: r6752]]

[[Imported from SVN: r6751]]

[[Imported from SVN: r6750]]

[[Imported from SVN: r6749]]

[[Imported from SVN: r6748]]

[[Imported from SVN: r6747]]

[[Imported from SVN: r6746]]

[[Imported from SVN: r6745]]

[[Imported from SVN: r6744]]

[[Imported from SVN: r6743]]

[[Imported from SVN: r6742]]

[[Imported from SVN: r6741]]

[[Imported from SVN: r6740]]

I'll fix that soon.  I want to take the time and see how
much the behavior changes.

[[Imported from SVN: r6739]]

[[Imported from SVN: r6738]]

[[Imported from SVN: r6737]]

[[Imported from SVN: r6736]]

[[Imported from SVN: r6735]]

[[Imported from SVN: r6734]]

[[Imported from SVN: r6722]]

[[Imported from SVN: r6721]]

[[Imported from SVN: r6720]]

make a new subdirectory 'coupling' for coupling stuff and move the new file rodcontinuumcomplex.hh there

[[Imported from SVN: r6719]]

[[Imported from SVN: r6718]]

[[Imported from SVN: r6717]]

This class should centralize some of the logistics needed for the knee
models and similar stuff.  Eventually, domain decomposition solvers should
be able to work on this data structure.  That is still a way off, though.

[[Imported from SVN: r6716]]

[[Imported from SVN: r6715]]

[[Imported from SVN: r6714]]