CAD score computation requires residue numbers in model and reference
to match. This raises an error in the Scorer object otherwise.
Catch this early on and raise an error specific to compare-structures
action with instructions on how to get rid of it.

Mainly for reporting in the compare-structures action
residues are reported as strings in format:
<cname>.<rnum>.<ins_code>, atoms are reported as
<cname>.<rnum>.<ins_code>.<aname>. This changes behaviour of
compare-structures and ost.mol.alg.stereochemistry
(ToJSON functions of violation info objects for the latter).

A global chain mapping can now be enabled with --global-chain-mapping as
a complement to the existing local chain mapping. This can be evaluated
for the scoring paper.

An ins code in the search ResNum caused a jump to the next residue.
The assertion that should've catched this is optimized away with -O3.
The new behaviour returns an invalid residue if the exact ResNum
(including ins code) is not present in these cases.

Calls the currently preferred chain mapping method that should
deal with most chain mapping cases in a reasonable runtime.

GetRepr identifies the best possible representation of a reference
substructre in a model including chain mapping etc. It also enriches
it with attributes/scores that are useful to analyse binding sites etc.
Enforcing a global mapping fully defines the desired representation in
the model but extracts it and enriches it with all the nice
attributes of the GetRepr output.

The wrong file was committed in @fa25403e.

This should have zero effect but you never know...

testutils.SetDefaultCompoundLib no longer needs to set the default
compound lib explicitly, as this is handled by import ost.
The method is still needed to check if a compound lib is available, and
was accordingly renamed to DefaultCompoundLibIsSet.

This sets up the compound lib upon calling import ost.
It simplifies the ost_startup.py script accordingly.
It is now possible to override the compound lib with
an environment variable: OST_COMPOUNDS_CHEMLIB

Previously, a new chain got added at the first newline after $$$$.
So if an sdf file ends with a newline, you got an unwanted empty chain.

That's the TMscore when using the other structure as reference