Add test for Median trait

[rustaceanrob]

The median time past check is failing over Testnet4 (issue #265). Here I am adding a check to be sure the Median trait defined in the prelude works as expected

cc @nyonson

[nyonson]

Do these fixtures need to be mutable?

[nyonson]

Ah, just found the median trait and see that it needs to sort.

[rustaceanrob]

Yeah, for now there is a sort, but it potentially makes the trait harder to reason about. In a follow-up it might be worthwhile to change this to a copy. Especially if it squashes the testnet4 bug

[nyonson]

Just spit balling, but I wonder if its worth considering a trait bound implementation (got this set of traits from Claude) to DRY-ify the impl.

use std::ops::Add;
use std::cmp::Ord;

pub trait Median {
    type Output;
    fn median(&mut self) -> Self::Output;
}

impl<T> Median for Vec<T>
where
    T: Copy + Ord + Add<Output = T> + From<u8>,
{
    type Output = T;

    fn median(&mut self) -> T {
        if self.is_empty() {
            return T::from(0);
        }

        self.sort();
        let len = self.len();
        let mid = len / 2;

        if len % 2 == 1 {
            self[mid]
        } else {
            (self[mid - 1] + self[mid]) / T::from(2)
        }
    }
}

Also, I believe there are some potential overflow cases? Haven't actually put the thought in though if the domain cares.

[rustaceanrob]

Skeptical of the From<u8> one. i.e. signed integers can't implement the trait. Otherwise, I like that it becomes generic over output. Worthwhile to revisit in another PR.

edit: Nevermind on From<u8>, that requirement is for the T::from(2) I see, which is no problem for signed types. Definitely worth the revisit then.

[nyonson]

ACK 037e262

Makes sense, sparked a few follow up ideas for later.