Port the most recent version of Musl's sqrt as a generic algorithm

Musl commit 97e9b73d59 ("math: new software sqrt") adds a new algorithm
using Goldschmidt division. Port this algorithm to Rust and make it
generic, which shows a notable performance improvement over the existing
algorithm.

This also allows adding square root routines for `f16` and `f128`.

Author: tgross35

etc/function-definitions.json

@@ -704,6 +704,7 @@
             "src/libm_helper.rs",
             "src/math/arch/i686.rs",
             "src/math/arch/wasm32.rs",
+            "src/math/generic/sqrt.rs",
             "src/math/sqrt.rs"
         ],
         "type": "f64"
@@ -712,6 +713,7 @@
         "sources": [
             "src/math/arch/i686.rs",
             "src/math/arch/wasm32.rs",
+            "src/math/generic/sqrt.rs",
             "src/math/sqrtf.rs"
         ],
         "type": "f32"

src/math/generic/mod.rs

@@ -1,9 +1,11 @@
 mod copysign;
 mod fabs;
 mod fdim;
+mod sqrt;
 mod trunc;
 
 pub use copysign::copysign;
 pub use fabs::fabs;
 pub use fdim::fdim;
+pub use sqrt::sqrt;
 pub use trunc::trunc;