Make `use_author_fields` default to `True`

[padix-key]

PDBx files offer to 'variants' of chain and residue IDs: the auth and label columns. Currently by default the auth columns are used for creating an AtomArray, to keep compatibility with the PDB format, which also uses the author annotations. However the label fields are much more consistent and are referenced throughout different PDBx categories.

As the PDB format is now deprecated a long time, I propose to change the default to label columns at the cost of consistency with the PDB.

However, one issue remains: Hetero residues that are outside a sequence have a label_seq_id of .. If in addition two of those subsequent residues have also the same residue name and chain ID, the residue-based functionality of Biotite cannot distinguish them as two separate residues (see also #553). So preferably this problem should be tackled first before moving to label fields.

Furthermore, auxiliary functions that use chain IDs should point them to label_asym_id instead.

So in summary, these are the working items:

• Fix missing bonds if use_author_fields=False when two same CCD together in wwPDB .cif file #553
• Use use_author_fields=True in get_structure() and get_assembly()
• Use label_asym_id as keys in the return values of get_sequence and get_sse()

[Croydon-Brixton]

What do you think of using the label res id's where they are available and default to the author res id's where they are not? This would handle the hetero residues with . annotation, making sure distinct residues do get distinct numbers without having to do any complex processing. Alternatively we could figure out residue boundaries from the author names for those and renumber ligand chains starting from 0/1?

[padix-key]

Alternatively we could figure out residue boundaries from the author names for those and renumber ligand chains starting from 0/1?

Especially this idea sounds quite intriguing to me. Currently, I cannot imagine an edge case where this would break something. And just counting up from 0 would also be consistent with e.g. the sym_id annotation.

[Croydon-Brixton]

Ok that sounds good to me - let's do that, this will also be a nice, consistent policy