xcbgen-parser: support switch-case

xcbgen-parser: support switch-case

"case" is implemented as a variation of "bitcase" as follows:

The old class "BitCaseType" became the abstract class
"CaseOrBitcaseType" with two derived classes
"CaseType" and "BitcaseType".
(Therefore BitcaseType keeps the old semantic which may
be important for some generators)

There are two additional flags in class Type:
* is_case
* is_case_or_bitcase
The latter is for convenience because case and bitcase
have to be treated the same way in many cases.

The list of cases and bitcases in a SwitchType object
is still called "bitcases".
Reasons:
* Renaming that list would break generators based on the parser.
* Creating an extra list for cases would make other code more
  complicated because you usually want to iterate over all
  case and bitcase members of a switch.

Reviewed-by: Ran Benita <ran234@gmail.com>