Add Inf and NaN literals to the DSL

[johnkerl]

This is in support of #908

You can already get IEEE Inf and NaN in the DSL by 1/0 and 0/0, respectively. The change is that now you can refer to them by name.

Also note that NaN != NaN -- this is a general face of IEEE floating-point arithmetic, nothing to do with Miller. So, now (like other languages) Miller has an is_nan DSL function to check if something is the not-a-number value.

$ rlwrap mlr repl
Miller 6.0.0-dev REPL for darwin/amd64/go1.17
Docs: https://miller.readthedocs.io
Type ':h' or ':help' for online help; ':q' or ':quit' to quit.
[mlr] 1/0
+Inf
[mlr] 0/0
NaN

[mlr] 1/0 == Inf
true
[mlr] 0/0 == NaN
false
[mlr] is_nan(0/0)
true
[mlr] is_nan(NaN)
true

[mlr] ?is_nan
is_nan  (class=typing #args=1) True if the argument is the NaN (not-a-number) floating-point value. Note that NaN has the property that NaN != NaN, so you need 'is_nan(x)' rather than 'x == NaN'.

[mlr] log10(0)
-Inf
[mlr] log10(-2)
NaN
[mlr] is_nan(log10(-2))
true