Simplify, permit user control over affinity

This ensures we 'do the right thing' when the user imposes affinity through e.g. Intel MPI's I_MPI_PIN_DOMAIN or other mechanisms. Also no longer does works with a shuffled ```cpuid``` array, which will hopefully have less surprising failure modes, and means we don't have to handle hyperthreading explicitly. If we want to be clever in this way e.g. to maximise available cache, then we could replace libnuma with hwloc as discussed elsewhere.

Hopefully this gives us a reasonable affinity which is easily overridden using standard techniques.

See merge request !120

Added name of parameter file to error messages.



See merge request !170

Task plot improvements

I have updated the plotting script and the output of the task data such that we can now have the full information for a given time-step and not just the last call to `engine_launch()`. This means the `drift` tasks will now be included in the plots. They will also include the single-core part where the cells and tasks are reconstructed. 
The start and end time as well as the CPU frequency are dumped in order for this to work.

Also, added a cosmetic change to the output for the case where a snapshot gets dump. 

See merge request !166

Better output of wallclock time when writing a snapshot. Slight improvement to the task plotting scripts

in the cpuid map. These are now offered in numa node order rather
than core order. This means that hyperthreads should be picked last.

With this fix we look to work well with Intel MPI using multiple ranks
on the same workstation, but PLATFORM MPI remains uncooperative.
To help with that a new command-line option -a has been added to
disable processor affinity, even when the numa policy is set
and available.

Better names for functions

This closes #143.

I have also updated the names of the functions in runner.c to stop linguistic problems between "dog host" and do ghost" for instance. 

See merge request !165

Time-step synchronisation

Ok, this is the final fix to #93. Particles are not allowed to move to a longer time-step unless we are currently doing a time-step of that size. That will prevent particles from being completely all over the place in the time-step hierarchy.

It wasn't physically wrong until now, just not efficient and very confusing.

With this fix in place, there will always be a time-step every so often where all particles are updated. 

See merge request !164

Conflicts:
        src/engine.c

Sph properties

Once !159 has been merged in, this merge request improves the handling of SPH parameters by actually reading them from the YAML file and not simply using the constants from const.h

See merge request !160

Timed outputs

Fix to #161.

Also corrects the run script for the external point mass that used to regenerate the ICs every time. 

See merge request !163