SireAdvanced Molecular Simulation Framework

I've added the VersionID.. type code to the DofID reading function. This fixes
the magic_error exception

Adding wrappers for DofID

Commit of the new python wrappers

Fixed a bug in the reading of MoverMove objects from a binary datastream

Increased the maximum number of alpha values for a softcore force field. I have done this so that I
could use different alpha parameters to soften serially the lennard jones and coulombic interactions
of a solute

I've committed the version of the code used to perform the simulations in the
WSRC paper (just submitted to J Chem Phys). This branch will be updated as
necessary to allow others to reproduce the simulations described in this
paper.

Looks like movermove is doing what I want on dimedone.

AngleDeltas were not working because I had forgotten to add the corresponding QHash to
SireMove_containers.cpp...

Added angle_deltas to MoverMove. Code appears to work for bond_deltas but not angle_deltas.

Commit of test job of movermove dimedone with dofid objects

Commit of the latest wrappers

Now working on adding bond_deltas and angle_deltas to MoverMove. Created a DofID class
to index the variable dofs of MoverMove. Code compiles but wrappers do not workx

I've changed the "at" function of ChunkedVector to remove bounds checking and
have switched the intermolecular forcefields over to using
ChunkedVector::at rather than ChunkedVector::operator[]. This has improved
the speed of the energy calculations by about 25-30%
(see test/SireMove/bigbox.py).

However, I am still performing a distance check between all pairs of
molecules, which means that I still have an O(n) cost in the
energy evaluation. This is painful, especially as this is evaluated
via a virtual function call (as it is Space dependent). I need
to think of some way of creating patches and breaking the system
down into patches (to remove the O(n) scaling).

I've fixed the bug in LJPotential that meant that recent versions of Sire
(which have not been used for any production calculations) calculated the
wrong LJ energy. It was a stupid error in LJFunction where I used
sigma^6/r^6 = pow_3(sigma/r^2) rather than the correct pow_3(sigma^2/r^2)

I've fixed a bug in RestraintFF that meant that it was not saving its "space"
property correctly. This caused errors to be thrown as RestraintFF forcefields
then had different spaces to non-RestraintFF forcefields.

I've added an example of how to create a simple harmonic restraint
between an atom and a point

I've fixed a bug in CovalentBondHunter whereby it never found any bonds
because it used a too small covalent radius - it now uses the bond order
radius. I've added an anchors example script to show how anchors are supposed
to work.

I am also trying now to speed up the code for large numbers of molecules,
in particular getting rid of some unnecessary order-N bits of code which
were seriously destroying the performance of the code for large water
boxes (hence the bigbox.py added example).

While this version compiles and runs, I wouldn't trust it yet - I've
done too much to the core and need to run a lot of tests.

Still working on movermove

Removing the debug comments...

I've added the code to allow the QM/MM potential to be used to polarise
the polarisable MM model