With this fix plumed crashed when using ADAPTIVE=DIFF and replica
exchange.

I think the way pbc were applied was not correct (Fix #19).

It was a longstanding (unknown) bug!
I added reference output files for all the simpledmd
regtest. This will allow to distinguish bugs in plumed
from Lyapunov effects in trajectories.

This fixes #23
Notice that it slows down (with icpc) generation of Gaussian numbers
by a factor 2.5. I think it is better now to have reproducible regtests.

added the keyword PERIODIC=NO in the example

Old patches have been remove.

I had to change the place plugin_forces is called in espresso.
This is because in the original implementation plugin_forces
was called *before* stress calculation.
I did some qualitative test, but I am not able to run a proper
thermostated variable cell simulation with QE. Probably will
require some extra validation by QE guys.

I also updated qe version to 5.0.2

Currently the .diff file is compulsory, the .config is not. So, the list
of MD codes should be built looking at .diff files

In PW, they use Ry units instead of Ha units, which leads
to a factor 2 in time. I realized it looking at the output of
espresso, where time is written also in ps. Now colvar and
output are consistent.

Virial was not scaled together with forces. The contribution is not
negligible, and I verified it is correct by testing with gromacs
a box of water in NPT and biasing the energy, then comparing the
-kt*log(histogram) biased and unbiased. It seems after this fix agreement
is much better

This closes #5

After Max validation I feel I can merge on release branch

There was a problem when running regtests using another plumed
executable (e.g. the one installed). In particular
a wrong warning was shown. This is fixed now.

Still we should write some doc. Also, there is not regtest for this...

This allows citing without explicitly including PlumedMain

* check that PLUMED_KERNEL is not redefined
* disable paths for libraries if plumed is statically compiled

Also fixed "make uninstall" so that it always clean up dynamic libraries

I moved the reference together with the others (using cite()) so that
it is more visible. I also updated a bit the description

This is the standard path for linux distribution. Thus "make install"
will fail unless run with root permission or unless PLUMED_PREFIX is
explicitly set. I also fixed DYLIB path for mac (plumed path
should be prepended)

In analysis runs results from previous analysis are saved as
analysis.0.<filename>, analysis.1.<filename> etc.  If the calculation
is NOT restarted and there are files with these names hanging about they are
renamed bck.0.analysis.0.<filename> bck.0.analysis.1.<filename> etc.

OptimalAlignment::weightedAlignment crashed with intel c++ -O3.
I was not able to find why (gdb and valgrind did not help).
With this workaround, that function only is optimized at -O2 anyway.

Fixes #7