Add NaN and Infinity Literal Syntax

[peblair]

See title. Open questions:

• What syntax should we use?
• Should they be F32 or F64? We should probably have both, since F32/F64 NaN comparisons may not work correctly (need to check)
• Perhaps others

[cician]

I think this is a duplicate of #360.

As a user I actually need to detect NaNs ind +/-infinite Numbers, but I think it needs to be done with functions like numberUtils.isNaN, which is for WASM primitives and not exposed.

[peblair]

@cician This is actually talking about the syntax, whereas #360 is referring to the constants. As for the isNaN and isFinite predicates, we should definitely add those. I think we should open a separate issue for that.

[cician]

OK, thank you. I've opened a separate issue.

[ospencer]

What syntax should we use?

I vote for NaN and Infinity.

Should they be F32 or F64? We should probably have both, since F32/F64 NaN comparisons may not work correctly (need to check)

Great question. Float literals of the Number type default to f64s, so I say we remain consistent. I think questions of equality should probably be handled by the operators.

As a side note, I would probably just make coerceFloat32ToNumber promote to an f64 and then simplify the Number operations to only worry about f64s and avoid the equality issue altogether.