Conflicts:
	.travis.yml

cppcheck is installed from its github when needed

Notice that the travis test doing cppcheck also does a normal g++ check.
I did to avoid too many if's in the travis script, but should be fixed perhaps
so that the last travis test only does cppcheck

I set this test as "allowed_failures", so that is just gives a warning now

I invoke cppcheck from an external script that makes the reporting better.

In particular, I show "fatal" errors at the end, where "fatal" means
all reported issues with the exclusion of "style" issues and issues
in molfile sourcecode.

(goes with 9e7b8c15)

Before this fix, the PLUMED_KERNEL env var was erroneously not
containing the LIBSUFFIX value.

This fix is relevant only if one use at the same time PLUMED_LIBSUFFIX
*and* loads the module file, which is unlikely.

Related to: e1b845b4

Now the type is defined with a typedef so that it is easier to change it.

I also set the type to size_t instead of unsigned long long.
This should be more portable.

removed condition that seemed to create very large support
in sum hills calculation and resulting into a very large computational
time. Credits to Carolyn Phillips from the mailing list.

gromacs-5.1.0 beta patch

[makedoc]

This reverts commit b148b715.

This should go with a94b94b2

[makedoc]

Similarly to what happens in Colvar class, I made
also for ActionWithVirtualAtom a method setBoxDerivativesNoPbc
that can be used to set the virial contribution when the Pbc have not
been used in the calculation of derivatives.

This should make the addition of new vatom actions easier

For DISTANCE, ANGLE, TORSION and PUCKERING,
I implemented something similar to GYR, where atoms coordinates
are first made whole, then the virial is computed assuming them whole.

Notice that modified coordinates are the local ones, so that the
change does not impact other actions.

Moreover, this required fixing numerical errors in a couple of regtests

Only add to the manual RESTART UPDATE_FROM and UPDATE_UNTIL
in the actions for which this makes sense

It seems the lapack routine I took from gromacs are not threadsafe.
Indeed, routines slasq5 and dlasq5 contained static variables.
I removed all of them and everything seems to still work correctly.
This seems to solve the problem that @gtribello
found in his no-buffer branch.

an INCLUDE, i cannot figure out if there was a reason for that

[makedoc]

Now all png needed for the manual are copied to html dir

[makedoc]