[makedoc]

wrong path for uninstall

Fixes #157

Clarification on one option that does not work on BG/Q

Lattice reduction is used in plumed to convert box to its
"least skewed" form. I changed a bit the threashold since it
was giving problems with high optimization flags.
Additionally, I added some warning when too many iterations
are performed (which indicates a numerical issue)

Conflicts:
	regtest/multicolvar/rt21/derivatives.reference
	regtest/multicolvar/rt21/plumed.dat

reduced precision to make them work with stronger optimizations

* added comments on install paths at the end of ./configure
* fixed include path in modulefile
* make obj/kernel.o only when necessary

Now configure properly takes into account paths such as
--bindir=/path
etc

This implementation hardcodes all the path into the
src/config/ConfigInstall.cpp file. I followed the instructions on autoconf
documentation so that the paths can be changed at install time using e.g.

make install prefix=/path bindir=/path/binxx

This should be the standard behavior.

I removed the possibility to use PLUMED_SUFFIX and PLUMED_LIBPREFIX
since it was a bit too difficult to maintain.

I significantly simplified the src/lib/Makefile by using
the standard directory names for autoconf.

Instead of replicating the whole tree upon install,
now objects are copied to prefix/lib/plumed/obj
and assigned a number.

When possible, objects are grouped in a single kernel.o
with ld -r (tested in configure). This saves a lot
of disk space.

I also removed from the install directory some other
useless files.

The Plumed.inc/cmake files when installing
are generated in a different way from the ones
used prior to installation. Thus, to avoid too many
repetitions, I wrote a small awk script the transforms
Plumed.inc to Plumed.make

Now when plumed is installed the scripts are placed in
lib/plumed/
instead of
bin/

This makes it more robust the installation of several versions with suffixes

I only left in place the script "plumed-patch" which is likely
the most used.

I added plenty of messages at the end of ./configure and make install
+ I explained it in the user doc

[makedoc]

[makedoc]

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

Properly instantiate templates when using high optimization flags
with gcc

Get_count() should return the number of element
only when used on a Recv status. In this regtests
we were probing the value of a Isend status.

Likely openmpi sets it anyway as the number of
sent element, but this makes mpich2 fail the regtest.

I thus removed the check.

It seems in 865d16c9 I introduced a bug:
checks where reported to travis only when
building doc.

Fixed now

[makedoc]

this is required to avoid name clashes

Now command line shortcuts such as plumed-patch
follow the same rule as plumed executable to set PLUMED_ROOT
as it was implemented in 1b20cb90

Namely, they have an hard coded value that can be replaced
setting the PLUMED_ROOT variable before launching the tool

Sometime there was problem with inconsistent on/off files. Now:

whenever all/none/reset is found, all on/off files are removed.
whenever one module is enabled, on file is added and off file is removed
the opposite when it is disabled

this should enforce consistency of on/off files.

Conflicts:
	src/maketools/update-plumedroot.sh
	user-doc/Installation.txt

Using "_" instead of "-" to avoid clashes.

Actually, the problem was that doing
make install PLUMED_SUFFIX=2
make install
The second install would have canceled the executables installed
by the first install. This is because install assumes names such as
plumed-* to be scripts installed for cross compiling

[makedoc]

There was an error in passing masses from NAMD to PLUMED.

Found by Glen Hocky.

Fixes #162

Now plumed should work if built on a directory which has a space in the path.
Notice that install directory should still not contain spaces.
This is because of some difficulties in making the
src/lib/Makefile rules compatible with names with spaces.

Addresses #157

to doxygen 1.8.7