Nios II floating point custom instructions

I designed a set of Nios II custom instructions implementing:

- single precision floating point add/sub, mul and div

- double precision floating point add/sub and div

together with the relevant C routines (__addsf3, __subsf3, __mulsf3, __divsf3, __adddf3, __subdf3, __divdf3) that replace the corresponding routines in libgcc.a.

The zip file also includes an implementation example targeting the Altera Nios II Stratix dev board.