Prototype implementation of Sage wrappers for FLINT generic rings

TODO:

(see also comments in code)

- more coercions and constructions
- more flexible handling of T_UNKNOWN
- inline element structs in Element object for at least some contexts
  (Element subclasses?)
- systematically wrap gr API
- more conversions from/to usual Sage objects
  (including fexpr <-> SR?)
- FlintPolynomial class (as subclass of FlintElement or Polynomial?)
- FlintMatrix
- FlintMPolynomial
- element subclasses with type-specific methods where relevant
- adapters for viewing Sage objects as Flint generic ring elements
- documentation, tests
- ...