throw an exceptio if an invalid preconditioner is used with steklovPoincare
more verbose with area debugging
added a setDataFromGetPot in steklovPoincare to setup aitken data
and preconditioner

minor

and don't do exit(1) that is very bad manner to quit just because
something went wrong, instead throw an exception

now check that the FSI solvers give the same results

Now ressources are properly allocated, used and destroyed
Alas the code is still fragile but way better

Fixed bug in steklovPoincare with the BC number 0 was picked to compute
quantities on the interface. Basically the order to enter the BC was important
now it is not a problem anymore

huge cleanup to avoid duplicating code. steklovPoincare and operFS are now smaller !

AM is porting her code to new FSISolver

of BCs

that's bad

use shared-ptr instead
now ressources are allocated and destroyed at the proper time

now the code is way more rebust and less error prone

now it takes a shared_ptr<DofInterfaceBase>

Vector --> VectorType

any other solver !

This is FSI made easy

It can/must be much improved but let's see that after 0.5

* added FSIFactory so that we have to manipulate only the base class operFS and
never the sublasses see test_fsi and the FSISolver class

* allow delayed complete initialization, because the factory will call the default constructor

 * Fixed a very nasty bug if we use a different FE space for the
   second domain than the one used by the first domain

 * Modified the class so that we can delay the real initialization

Uses now Gilles interface classes for FSI. There is not yet an interface class for FSI-0d coupling due to some technical problems... Think it will be for release 0.6!

make distchekc runs properly from beginning to end

lifev 0.5 is almost ready for release :)

Gilles should add test_fsi to testsuite very soon