Fof documentation

Add documentation for the FOF code:

• General description
• Stand-alone mode
• On-the-fly + BH seeding
• Parameters

@jwillis can you let me know whether this all makes sense and is correct? And that nothing important is missing? I am planning to add links to your paper and possibly a figure from it once published.

doc/RTD/source/FriendsOfFriends/algorithm_description.rst

+.. Friends Of Friends
+   Matthieu Schaller 15th June 2019
+
+.. _fof_algorithm_description_label:
+
+Friends-Of-Friends Algorithm
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Friends-Of-Friends (FOF) is a common tool used to identify groups of
+particles in a simulation. This can, for instance, be used to identify
+haloes in a cosmological simulation.
+
+In practice, this is done by using a *linking length* ``l`` and
+demanding that any particle that finds another particle within a
+distance ``l`` is linked to it to form a group. A particle is linked
+directly to all other particles within a distance ``l`` (its
+*friends*) and indirectly to all particles that are linked to its
+friends (its *friends-of-friends*). This creates networks of linked particles
+which are called *groups*. The size (or length) of
+a group is the number of particles in that group. If a particle does not
+find any other particle within ``l`` then it forms its own group of
+size 1. For a given distribution of particles the resulting list of
+groups is unique and unambiguously defined.
+
+Small groups are typically discarded, the final catalogue only contains
+objects with a length above a minimal threshold, typically of the
+order of ``20`` particles. Smaller groups can often be spurious.
+
+Once the groups have been identified, properties can be computed for
+each of them. The total mass or the centre of mass are common
+examples. These are then stored in catalogues alongside a unique
+identifier for each group.
+
+SWIFT implements FOF using a Union-Find approach. It also exploits the
+domain decomposition and tree structure that is created for the other
+parts of the code. The tree can be easily used to find neighbours of
+particles within the linking length.
+
+Depending on the application, the choice of linking length and
+minimal group size can vary. For cosmological applications, bound
+structures (dark matter haloes) are traditionally identified using a
+linking length expressed as :math:`0.2` of the mean inter-particle
+separation :math:`d` in the simulation which is given by :math:`d =
+\sqrt[3]{\frac{V}{N}}`, where :math:`N` is the number of particles in
+the simulation and :math:`V` is the simulation (co-moving)
+volume. Usually only dark matter particles are considered for the
+number :math:`N`. Other particle types are linked but do not
+participate in the calculation of the linking length. Experience shows
+that this produces groups that are similar to the commonly adopted
+(but much more complex) definition of virialised haloes. A minimal
+group length of :math:`32` is often adopted in order to get a robust
+catalogue of haloes and compute a good halo mass function.
+
+For non-cosmological applications of the FOF algorithm, the choice of
+the linking length is more difficult and left to the user. The choice
+of the minimal group size to consider is also application dependent.
+
+In SWIFT, FOF is also used to identify haloes in which to seed black
+holes in a cosmological run. When used in this mode, the code does not
+produce any outputs but uses the catalogue internally to convert some
+gas particles into black hole particles.
+