Ihm

=================
- BinaryCIF files now use UTF8 msgpack strings for all text, rather than
raw bytes. This should make python-ihm's BinaryCIF files interoperable
with those used by, e.g., CoordinateServer.
- Output mmCIF files now include author-provided numbering (auth_seq_id)
for atoms in the atom_site table. This should help packages that don't
read the pdbx_poly_seq_scheme table to show the desired residue
numbering (61).

=================
- Support for Python 2.6 has been dropped. The library needs Python 2.7
or Python 3.
- Bugfix: correctly read in multiline reference sequence one-letter codes.
- Bugfix: the reader is now more tolerant of omitted or unknown values
(. or ?) in input mmCIF files.

=================
- A convenience class is added to describe datasets stored in the
ProXL database (:class:`ihm.location.ProXLLocation`).

=================
- Update to match latest FLR dictionary.
- Add a simple utility (util/make-mmcif.py) to make a minimal compliant
IHM mmCIF file, given an mmCIF file (potentially just coordinates) as input.
- Bugfix: the full residue range spanned by a starting model is now reported,
rather than just the subset that is mapped to one or more templates (55).
- Bugfix: handle TrEMBL UniProt sequences (57).

=================
- Convenience classes are added to describe hydrogen/deuterium exchange
data (:class:`ihm.dataset.HDXDataset`) and datasets stored in the
PDB-Dev database (:class:`ihm.location.PDBDevLocation`).
- Multiple :class:`ihm.restraint.CrossLinkPseudoSite` objects can now
be assigned to a given :class:`ihm.restraint.CrossLink`.
- Bugfix: the :class:`ihm.dataset.Dataset` base class now has a type
of "Other" rather than "unspecified" to conform with the latest
IHM dictionary.

=================
- :func:`ihm.reader.read` no longer discards models read from non-IHM mmCIF
files; they are instead placed in their own :class:`ihm.model.ModelGroup`.
- Bugfix: both the pure Python and C-accelerated mmCIF readers are now more
robust, able to handle files in binary mode (e.g. from opening a URL)
and in Unicode (mmCIF files are supposed to be ASCII but python-ihm should
handle any encoding Python supports).