When using METAD with file-based multiple walkers and many processes
launched with mpirun (e.g. with gromacs -multi) replica suffix is
not added to hills file.

Contributed by Marco De La Pierre

gaussian with grids, the check for sigma_min and
grid was not working properly

the gaussian support was wrong

[makedoc]

now METAD and PBMETAD can write grid files upon checkpointing
related to #37 and #125

-the component bias and a pointer to it are now initialised in Bias.cpp
-the derivatives for bias are automatically set from forces  in apply

- METAD: exit with error if used
- PBMETAD: gives a warning and deactivate the interval for the specific variable
close #181

[makedoc]

close #21

- added the restart from grid to PBMETAD
- more error checking

- when restart from GRID (RGRID+RESTART) doens't read the HILLS (but continue to append)
- if not restart then the HILLS are added to the existing GRID
- the grid is not external anymore, in this way when you write the grid you write the total one
- grid files are written only once in case of multiple walkers
NOTE: the grid is not modified for the BIASFACTOR because the BIASFACTOR is not currently stored
in the grid, this means that if you restart from a grid

- reformatted
- restart from a grid doens't read HILLS
- in MW grid are only written once as for HILLS

I cleaned up some of the (innocuous) warnings

this check should be done also in NDEBUG mode

turnOnDerivatives() was missing at the end of MetaD

This is required because of the limitations of the previous commit

Same as #171 but using C++ standard include file <ctime>

Closes #171

This bug was introduced by me in 5e3aaf3a
The barrier on multi replicas was due to overcaution and
is only necessary when using WALKERS_MPI

It seems that sometime on travis a regtest with mw-mpi crashes.

I think this is because when using multiple processors per walker
there is a problem in the opening of the old file (for read) and
its backup (for write). Adding this barrier seems to make it more robust

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.