Simplify implementation

[mgeier]

This PR contains two orthogonal changes:

• the code is simplified by introducing an additional type definition per operator
• the impls themselves are simplified

I combined the two because the second is easier to do in the simplified code after the first change.

I found out that the impls can be simplified by reading https://github.com/willcrichton/tyrade/blob/master/src/tnum.rs. I extended that idea to the other operators as well.

[fprasx]

I'm not looking to change main.rs, but if you'd like to add this implementation in another file (with proper attribution to yourself and willcrichton), I'm happy to merge.

[mgeier]

I'm not looking to change main.rs

No problem, don't hesitate to close this PR without merging.

I just wanted to preserve this for my own future reference, for which this PR is sufficient.

---

For the record:

This reduces the "monstrosity" mentioned in https://fprasx.github.io/articles/type-system-arithmetic/ to:

type RawQuotient<T, U> = Add<Succ<Zero>, Div<Sub<Succ<T>, Succ<U>>, Succ<U>>>;

impl<T, U> DivImpl for (Succ<T>, Succ<U>)
where
    (T, U): SubImpl,
    (Sub<T, U>, Succ<U>): DivImpl,
    (Succ<Zero>, Div<Sub<T, U>, Succ<U>>): AddImpl,
    (
        GreaterThanEq<T, U>,
        Add<Succ<Zero>, Div<Sub<T, U>, Succ<U>>>,
    ): MulImpl,
    (T, U): GreaterThanEqImpl,
{
    // If x < y, return 0. We can do this by multiplying the "RawQuotient" by
    // the bool->int value of this condition.
    type Output = Mul<GreaterThanEq<Succ<T>, Succ<U>>, RawQuotient<T, U>>;
}

It's still quite a mouthful, but I think better than before and I think it is now short enough to inline RawQuotient, which unnecessarily confused me when first reading the blog post.

A more concise implementation is available at fprasx/fprasx.github.io#1 (comment).