Make the code generic under num_traits::Float?

[grothesque]

Wouldn’t it make sense to make the algorithms of this library generic under the num_traits::Float trait?

Then, one could not only use double-f64, but also double-f32, and any other type that implements that trait, e.g. the ones provided by the half crate, or future f16 and f128 in Rust.

[shinaoka]

Agreed. I have requested this, but Markus is away.

[mwallerb]

@grothesque
Yes, that makes a lot of sense!

I am not sure which algorithms you mean, but the ones under ::arith are good candidates for making them generic, for two reasons: one, as you say one could do double-f32, but more importantly, it would make using SIMD very easy.