The result are valid queries of chains with name '_'.

Fixes the error "ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()" in AddMesh.

Atoms that do not actually exist are still added in deep view. The problem is
that they're added at position (9999.99, 9999.99, 9999.99). Internally we use
a grid of equally sized cubes to efficiently lookup up positions closeby.
Extreme locations lead to an extremely large grid with most of the cubes being
unoccupied. The simple solution here is to just increase the size of the
cubes. The larger the cubes, the slower the code. However, in real world
scenarios their size won't be inreased.

hasn't been used/tested for years

The required functionality has been removed in commit:
41ba5333

Use the WrappedTMAlign instead

Image export has been performed using platform dependent offscreen buffers
that are setup manually. Segfaults etc. were observed on several occasions.
This commit enables offscreen rendering through the GLCanvas using QT
functionality.
Pro:
- QT takes care of platform dependent things
- Other image formats than just png come for free
Con:
- The gfx module didn't depend on QT and image export could be scripted
  Scripting is still possible, but a GLCanvas must be created.
- A offscreen solution could be implemented but it would still depend on QT

Ignore missing (struct_conf|struct_sheet_range).(beg|end)_label_seq_id data item in fault_tolerant mode of mmCIF reader.