k256+primeorder: add BatchInvert and BatchNormalize impls

[ycscaly]

Addresses #943

[ycscaly]

@tarcieri there are the tests for inversion, and also implemented the projective to affine batch conversion.
However, this is only for k256 for now. I can do it for other crates, but its a bit of repetitive work - perhaps we leave it so that if any community request for it comes up, they can copy my logic?

[tarcieri]

However, this is only for k256 for now. I can do it for other crates, but its a bit of repetitive work

Once you have k256 nailed down, the only other thing that needs it is primeorder, then all of the other crates will benefit.

[ycscaly]

However, this is only for k256 for now. I can do it for other crates, but its a bit of repetitive work

Once you have k256 nailed down, the only other thing that needs it is primeorder, then all of the other crates will benefit.

Ok ok, I did it

[ycscaly]

@tarcieri should be good now

[tarcieri]

Why did you remove all of the Invert impls on scalars?

[ycscaly]

Why did you remove all of the Invert impls on scalars?

Because I added a macro that implements it, so they were double implementations

[tarcieri]

In places where you reference alloc::vec::Vec, particularly the projective.rs files, I think the code would be a bit less noisy with a conditional import:

#[cfg(feature = "alloc")]
use alloc::vec::Vec;

[ycscaly]

Thanks!