@alexvoronov pointed out that the example was actually wrong.

No EOL at the end of the file is only tolerated in input file
(e.g. plumed.dat) and not in other files (including HILLS files).

Indeed, when using multiple walkers, it is possible that transiently
the HILLS file from other walkers have incomplete lines.
With this fix, those lines are ignored and read at the next iteration.

This fixes #190

Prvented division by zero in calculation of average field and added TRANSFORM analogues of multicolvar filters

- faster calculation for RADIUS and TRACE
- default is now NOT_MASS_WEIGHTED
- regtest updated

Gaussian kernel functions were normalised incorrectly.  This should
only cause a multiplicative scaling of the histograms calculated
using the HISTOGRAM action

Changed from left to right handed axis system in MultiColvarDens and now doing proper spatial average of multicolvar values

Added option in driver to evaluate forces due to a bias numerically so that they can be compared with the analytical forces evaluated by plumed

- Implementation of short help (with :PHelp)
- Updated doc
- Autocompletion now knows if label was already provided at the beginning
  of the line and does not suggest it anymore in case
- Compilation of the vim plugin is more similar to compilation
  of the user doc
- Compilation of the vim plugin is made when doing "make install"
  from root dir.

- Made plumed dictionaries local to buffer
- Improved syntax annotation
  Now it also annotates spaces. Moreover, output is formatted.
  (requires a rt to be reset)
- Fixed nested parenthesis
- Made all regions foldable (include {})

also for action where not all the components have
derivatives

[makedoc]

allows syntax to be checked