This is to avoid false errors in regtest

Coordination is now faster by approx 15% when using even
NN, even MM and d0=0

Closes #43.

Conflicts:
	src/core/CLToolMain.cpp

This allows them to be automatically skipped when molfile has
not been linked

This minimizes the difference between code with __HAS_MOLFILE
and code without it

@ sign is now used within plumed to specify special groups of
atoms or special colvars whose labels are created by plumed and
not by a user

The source code from crystallization allows you to calculate
Q4 and Q6 parameters.  You can also calculate local Q4 and local
Q6 parameters and you create compound collective variables based
on the relative orientations of internal vectors in adjacent molecules.

Regtests for this functionality are reasonably thorough.  Manual is
not so great as yet

Conflicts:
	user-doc/Intro.txt

I removed one page (WhatsNew) and merged it in the changelog page for
2.0

This makes it such the order in which dependencies are activated
is deterministic. It is related to a bug (#51) which was already fixed
by changing the way prepare() works in multicolvar. Anyway, I commit
this change because I think it is clearer if, in a multiple proc simulation,
all the procs prepare all the actions in the very same order.

This addresses bug #51.  I think there can potentially be problems
if you call mpi functions from prepare I don't fully understand why
perhaps @GiovanniBussi has some understanding.

Plumed paper has now a volume and page

factor as proposed in Tiwary and Parrinello, PRL (in press)

Plumed.inc and Plumed.cmake are now regenerated each time.

(cherry picked from commit cff67707)

Plumed.inc and Plumed.cmake are now regenerated each time.