Fixes to postprocessing: Remove unused code, test vector extraction, unify treatment of scalar fields

• Remove the unused FlattenSequences class from ir.postprocessing
• Add a test case for vector extraction
• Treat explicitly and implicitly modelled scalar fields the same in map_field

Concerning Scalar Fields

pystencils permits users to omit the trivial index shape (1,) for scalar fields. We distinguish D+1-dimensional explicit scalar fields, where the trivial D+1st dimension is modelled, and equivalent D-dimensional implicit scalar fields. To make sure that explicit scalar fields can always be mapped onto D-dimensional data structures, we ignore their trivial index dimension.