Extend floating point extension with isnan, isnf, and integer conversions

[sampsyo]

I am also interested in the following:

If there should be 2 additional operators to the floating point extension, isnan and isinf, which would match similar behavior as seen in Python. Though I guess that this could be implemented as a small function calls in a Bril program and then used via the import extension in the niche cases that would want them.

If Bril should allow conversion between integers and floating point values with an operator like toint/tofloat.

Originally posted by @Pat-Lafon in #214 (comment)