This way we avoid making networkx a hard dependency on mol.alg.

Can be considered backup commit and still contains debug output

This is no final product but just a first working version as a
starting point. Plenty of testing/cleanup/speedup required.

This indicates a bug that must be fixed but at least we don't run
forever.

Use a very large inclusion radius. This should not impact the exhaustivity
of the chain mapping search by much, but makes it less likely to impact
low/medium RMSD scores.

These are two independent ways to speed up the processing but they
make no sense in combination and could lead to unexpected effects.

- separate scrmsd and lddtpli computations in separate functions
- optionally reduce the binding site search space in the model

This should not change behavior but will make it easier to speedup
binding site extraction

LigandScorer uses ChainMapper.GetRepr to match binding sites between target
and model. This commit limits the search space for model binding sites to
locations where there actually is a ligand.

An lddt_pli_radius < radius may lead to the absence of contacts in
lDDT-PLI.

When radius > lddt_pli_radius, it can happen that the lDDT value for
lDDT-PLI is undefined because there are no contacts, and is set to None
by definition. We handle it by not assigning the ligand (for lddt_pli
only) and add a new "unassigned" reason.

Special characters in chain names, such as '.' will break things otherwise

It turns out for some reason we were reading them from chains, based on
the ChainType of the chain. This prevented extracting anything from
legacy PDB files and didn't match the documented behavior in the action.

Now extraction is based on the IsLigand flag only, and documentation is
consistent throughout. Behavior on PDB files is tested.

Some ligands like YUT (in CAMEO 2023-12-30_00000023/8G9C) have a huge
number of isomorphic mappings and cause CAMEO to run out of time in the
scoring (and fail horribly). This allows to limit how much time is spent
listing all the isomorphisms. 100000 (1e5) seems reasonable and ensures
the scoring completes in minutes. Computing RMSD with SCRMSD is a bit
faster and 1000000 (1e6) should be OK.

New unassignment reasons are introduced as well.

Move long description to separate properties