Adds conversion functions from integers to floats

[Mottl]

This PR adds four conversion functions from integers to floats:
f32x4::from_i32x4
f32x8::from_i32x8
f64x2::from_i32x4
f64x4::from_i32x4

[Lokathor]

Looks like the build errors are because you're trying to use nightly stuff.

Also, these should either be From impls, or they should have more descriptive names that help the reader understand why it's not just a normal From impl. eg: f64x2::from_i32x4_lower2(x) lets the reader understand that they're dropping half their lanes.

[Mottl]

Thanks for comments! I chose to implement From.
Do you want to have f64x2::from_i32x4_lower2(x) instead of impl From<i32x4> for f64x2?

[Mottl]

Btw, is there a reason for aarch64 in #[cfg(all(target_feature="neon",target_arch="aarch64"))]?
(compare with #[cfg(target_feature="sse2")]).

[mcroomp]

They aren't really simple From since they are potentially lossy... i32 -> f32

[Mottl]

Yes, according to From description you are right. Changed from From impls to ordinary functions.

[Mottl]

@Lokathor, does it look good now?

[Lokathor]

Released wide-0.7.26

Sorry about the delay! last weekend got unexpectedly busy