1e0 / 0e0 should return Inf

[nwellnhof]

I'm seeing:

say 1e0 / 0e0  # Attempt to divide 1 by zero using /
say 1e0 % 0e0  # Attempt to divide 1 by zero using %

I'd expect to see Inf (as mandated by IEEE 754) and NaN (like POSIX fmod).

[nwellnhof]

I think that the DivideByZero failures in Num.p6 simply have to be removed.

[zoffixznet]

The current behaviour is part of 6.c specification, so the fix is more involved than simply removing those Failures. It has to be language-version based.

That was on 6.d roadmap, but is "blocked" by R#1289 and so was deferred until later language version. I use "blocked" in quotes because technically we can currently re-define these operators in 6.e.PREVIEW language already, but:

• We'd have to redefine all candidates, because otherwise we get dispatch conflict
• The main core code that uses the ops would still receive 6.c semantics

Thus, trying to make it work right now would at the very least be fragile and messy, so the hope is that once we figure out all the semantics to have R#1289 closed we will have figured out how to handle language-version control of these ops' behaviour as well.