Based on a CAMEO target (8TRK) with KFM ligand (82944 symmetries) which
takes about 15 minutes per ligand pair. Limiting to 10000 (1E4) should
limit to more reasonable 2 minutes / ligand pair..

LogDebug never shown in optimized builds

If it's not an SDF file then the error was meaningless. Make it
clearer that we may not have an SDF file at all

This commits makes the SDFReader aware of IOProfiles. It allows bad
bond types in fault_tolerant mode (ie dative bonds from RDKit with
non-standard type 9).

Whether we used bb_pos or full_bb_pos for superposition. Avoids
returning the RMSD of a single atom.

No one uses it (fingers crossed) and it clutters documentation with
unnecessary stuff

SCHWED-6400

Potentially triggers an error though

also fixes inconsistencies in scoring.Scorer doc strings

That means: apply chain mapping and then remove all model residues
for which there is no target counterpart. We therefore remove
model contacts for which we simply have no experimental evidence.

See announcement from PDB:
https://www.wwpdb.org/news/news?year=2023#65562f0ad78e004e766a96c1

The rmsd_ prefix was more confusing than helpful to Mike