parse_logic does not recognize QF_FPBV

[christophejunke]

Some benchmarks and solvers apparently support the syntax QF_FPBV (in that order) for QF_BVFP. For example, the following repository has a lot of examples of this:
https://github.com/florianschanda/smtlib_schanda

The code in typecheck/logic.ml seems to require BV to appear first:

dolmen/src/typecheck/logic.ml

Line 145 in ec32e34

and parse_bv c = function

I couldn't find an explicit rule about how logics should be named, but I guess that since this variation occurs in the wild, it might be preferable to support it if possible.

[Gbury]

That's a good question, I think i'll try and raise it on the SMT-LIB2 spec repo to see what the SMT-LIB maintainers think about that.

That being said, once I finish the work on "printers" (or rather the ability to export problems to the SMT-LIB format), I expect that the "good" solution will be to just ask dolmen to reformat the problem so that it becomes correct.