Add support for FP16 isFinite, isInfinite and isNaN

This patch adds mid end support in c2 for FP16 operations - isFinite, isInfinite and isNaN. It adds aarch64 backend support for these operations as well.