Add floating point support

[chsasank]

Add floating point types

https://capra.cs.cornell.edu/bril/lang/float.html

and tests:

https://github.com/sampsyo/bril/tree/main/test/interp/float

Subset of #13

[GlowingScrewdriver]

LLVM supports ordered and unordered floating point operations. BRIL doesn't make a distinction between them.
Whether a comparison or ordered or not affects the outcome when one operand is NaN. I am not sure of the finer details.

Clang emits ordered comparison instructions when comparing arbitrary numbers. We will follow the same for now.

[GlowingScrewdriver]

Okay. Looks like the difference is this: if either operand is NaN, ordered operations return false, while unordered operations return true. So, makes sense to support the ordered operations.