Possible bug with Either.Left short-circuiting

[Harmoniguy]

I might be misunderstanding the intention, but I think I have found a bug in Either.

My understanding to the Either type is that Left should always short-circuit, as is stated in the readme.
However, as an example:
Either.left('x').app(Either.left('y'))
will result in:
Either.left('y')
rather than what I would expect:
Either.left('x')

As a sanity check I wrote what I understood to be the equivalent in Haskell:
Left 'x' <*> Left 'y'
which results in:
Left 'x'

In an effort to improve test coverage, I have altered the EitherTest apply property test to test for this case:

    @Property
    public void apply(char c, char d) {
        assertEquals(Either.right(String.valueOf(c)), Either.right(c).app(Either.right(Object::toString)));
        assertEquals(Either.left("fail"), Either.right(c).app(Either.left("fail")));
        assertEquals(Either.left(c), Either.left(c).app(Either.right(Object::toString)));

        Assume.assumeTrue(c != d);
        assertEquals(Either.left(c), Either.left(c).app(Either.left(d)));
    }

Please advise whether this is the intention, or if you would like I can open a PR with the test.

[jon-hanson]

I think you're right - the Either.app implementation is bugged. I've committed a fix and have added your test (which now passes). Can you take a look.