This funcitonality allows you to add tutorials written in Doxygen
and links to external websites.  They are called How Tos though

It is now possible to use "make lib" also in root directory

Previously it was only possible to calculate the average value of the multicolvar
in a part of the box.  Now you can calculate the distribution inside the box.
This means you can use the functionality in ActionWithVessel to calculate the minimum,
the number of colvars less than, number of colvars more than and so on.

Small changes to core headers are changes to make certain routines virtual so that I
can replace them in derived classes

In multicolvar if you have a list of sets of atoms in each colvar
i.e ATOMS1=1,2 ATOMS2=3,4 ATOMS3=5,6 you can now replace the ATOMS
with another word and use the same functionality.

I want to reuse stuff from these classes to do some other things and
I needed to make a few small changes to make the new things work.

Mostly unused variables

The virial computed with NUMERICAL_DERIVATIVES was transposed
wrt to the analytical one. Fixed now.

It was not doing proper center-of-mass fitting. Now it also implements
the case with displace!=align. Corresponding regtest has been fixed,
and rt14 has been slightly changed (with new numbers it was not meaningful).

Without this change the code was compiling on RGPs local machines

Fixed a number of errors in the checks performed when vessels/multicolvars are compiled without the -DNDEBUG flag

It is automatically used whenever it is expected to produce
results identical to the old version. Thus, old version is still used
only when align!=displace or when not all align!=0 are identical

It is still possible to enforce non-safe (faster) version with
OPTIMAL-FAST.

TYPE=OPTIMAL is the old version
TYPE=OPTIMAL-FAST is the new version

Notice that the new version only works with align==displace.
Also notice that old/new version give inconsistent results when weights which are not all equal (or zero) are used. This is something that will be fixed
soon in the old version.

The new version is only used in a couple of regtests, since it still have some problem of numerical stability

RMSD re-implemented from scracth for the case where align==displace.
The new routine is much faster, but has some problem of numerical stability
(rt32 is broken). The code is choosing automatically between new version
(when align==displace) and old version (when align!=displace) so as
to allow the latter for e.g. docking. In RMSD.cpp, you can edit these lines:
// set to true one of these to enforce old routine (Davide's) or new one (Giovanni's)
  const bool enforce_old=false;
  const bool enforce_new=false;
to enforce old/new version.

Avoid clashes between same patch created by different people

Now the specialized forms (*.static/shared/runtime) only
contain the relevant definitions

Essentially added stride and off to chainRuleForElementDerivatives.
This should make it so that I can accumualte the values and cross
derivatives using this routine

It is very difficult to write a generic description of the GROUP keyword
so I changed the documentation so that you have to write documentation for
GROUP in the MultiColvar you are implementing.  So now DISTANCES contains
the keys.add(atoms-1,"GROUP") bussiness, this is no longer in MultiColvar

This means we can use MultiColvar to do things like the average angle in
the first hydration sphere or the distribution of angles in the first
coordination sphere.

This means you can create FunctionVessels with a specific name rather
than just the keyword

These tests were causing problems because of small numerical errors

There was an error (force out of grid requested).
Still, this regtest is not very clever (potential is 0).