Add float and integer traits from compiler-builtins

In preparation of adding generic algorithms to `libm`, add the traits
from `compiler-builtins`.

Eventually we should be able to unify the two crates so we don't have
duplicate implementations.

(cherry picked from commit 8241ad7)

Author: tgross35

src/math/mod.rs

@@ -323,6 +323,7 @@ mod k_tanf;
 mod rem_pio2;
 mod rem_pio2_large;
 mod rem_pio2f;
+mod traits;
 
 // Private re-imports
 use self::expo2::expo2;