This makes these options consistent with others.

THIS REQUIRES ./configure AGAIN!!

[makedoc]

I added the possibility to dump the 9 components of the virial
from the driver.

[makedoc]

This allow amber formats to be properly read

I added a new action DUMPMASSCHARGE and a new flag for driver --mc

Fix #112

Notice that I did it differently wrt what I wrote in the issue.
Namely, I write mass and charges with an action (DUMPMASSCHARGE)
and I make the driver capable of reading them (with --mc)

Since now plumed can write xtc/trr files, in a multi replica
framework I think it makes more sense to add the replica suffix before
the extension. That is, with DUMPATOMS FILE=test.xtc, we get
files such as test.0.xtc and not test.xtc.0

The same happens for trajectories read from the driver.

Notice that this was already true for .gz files, and now applies
for .xtc and .trr files. All other extensions are still ignored
for backward compatibility.

Now driver can read using --ixtc and --itrr with the xdrfile library.
This is an alternative to molfile, and provides better support
for cells of arbitrary shape.

Now all the types and functions declared in the
local version of molfile plugins are properly wrapped
in a namespace.

Possibly fixes #110

(cherry picked from commit 3f296491)

Changes to the code are there so that I can reference the atoms
page in the descriptions of keywords that read atom numbers.  I also
changed the input to multicolvar functions.  I stopped using the ARG
keyword in these actions and used DATA instead because you are refering
to the data calculated in a multicolvar as opposed to a scalar quantity
calculated by a particular action.

Sorry for this - on linux/intel it was compiling properly

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

I fixed the usage of stopflag in driver so that it is clearer
that the program stops at the end of the present step and not
at the beginning of the following. This is obviously equivalent,
but shows that scope of the flag can be reduced to be defined
at the inner of the MD loop.

Even if this is not a bux fix, I consider it as paired with fix #47
and I merge it to v2.0

I think this is more coherent with other cmd() calls.

I set it to 2 instead of 2.0, so that it will not be necessary
to update it

I replaced everywhere it made sense the new Communicator class
syntax. This commit is not strictly necessary, but provides
examples of the simpler syntax so I think should be merged

I adapted the debug replica exchange stuff. Since running on multiple
trajectory could be useful, I leave --multi as a non-hidden option.

I slightly changed behavior of the hidden flag --debug-grex (now one sets the exchange stride, not the number of replicas - the latter is set with --multi). Thus, a couple of retests had to be updated