Add support for floats

[AntonLydike]

Adding a Float32 datatype is necessary, since python makes no guarantees to the bitwidth of float (it's often a double)

Also adding the RV32F extension with most operations implemented, and support for floating point registers.

[AntonLydike]

Missing implementations:

• floating point classify instruction: fclass.s
• sign twiddlers: fsgnj.s, fsgnjn.s, fsgnjx.s

@superlopuh Do we need these or do you think we can merge with this?