[wpimath] Add Sleipnir Java bindings

[calcmogul]

The wrapper includes reverse mode autodiff, the Problem DSL, and the optimal control problem API. I wrote it by directly translating the upstream API and tests to Java (i.e., copy-paste-modify).

I replaced the ArmFeedforward and Ellipse2d JNIs with implementations using the Sleipnir Java bindings. Switching dev binary JNIs to release by default sped up wpimath test runs from several minutes to 7 seconds.

[KangarooKoala]

Still haven't gotten to the tests, but the source looks pretty good.

[KangarooKoala]

Given the size of this PR, it'd be nice if the sleipnir update had it's own PR so that its changes wouldn't be buried.

[calcmogul]

Yea, I was mainly bumping it here to keep it in sync with upstream bugfixes. I'll make a separate PR once review is done and the bugfixes slow down.

[KangarooKoala]

Still reviewing the other tests, but submitting some initial comments now so that those can get addressed.

On another note, the tests leak every single Variable (including intermediates) since they are never closed. Unfortunately, I don't think we can really do anything about that (at least, not on the test side).

[calcmogul]

Yea... I was hoping the GC called close(), but it seems not. In any case, manually closing them or using try-with-resources with all of them is too onerous. Java deprecated and removed finalizers in favor of try-with-resource and cleaners, but idk if cleaners are much better.

[KangarooKoala]

Done reviewing everything now (though a few parts were more verifying that it matched the original Sleipnir, not necessarily verifying correctness).

[KangarooKoala]

Seems fine to me for what it's worth, except some lingering concerns about the difference in accuracy of FlywheelOCP.

[KangarooKoala]

Now that the FlywheelOCP stuff is sorted out, seems fine to me for what it's worth.