Implement theory parsing

[namcsi]

It would be nice to be able to reify input theory atoms in a parsed format, such as in clingox.ast.TheoryParser, as this would make writing meta-encodings that work with theory atoms easier in some cases.

I see two options for implementing something like that:

• Just have clingox.ast.TheoryParser take a pass over the AST before reification.
• Implement the transformation the TheoryParser does using a meta-encoding via renopro.

The first option seems the wiser of the two. Should probably be provided as an option, in case someone for some reason would like to access the unparsed theory terms.