Add second-order derivative functions

[gerlero]

Draft PR for changes discussed in https://github.com/JuliaDiff/ForwardDiff.jl/pull/678

Fixes #71.

[codecov[bot]]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (211b675) 82.94% compared to head (72e4201) 83.37%. Report is 2 commits behind head on master.

Files	Patch %	Lines
src/AbstractDifferentiation.jl	93.33%	1 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #122 +/- ## ========================================== + Coverage 82.94% 83.37% +0.42% ========================================== Files 8 8 Lines 428 451 +23 ========================================== + Hits 355 376 +21 - Misses 73 75 +2 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[gdalle]

I think the change is welcome here and coherent regardless of whether it makes it into ForwardDiff. However it probably needs tests too, which I guess is why the PR remains a draft. Thank you so much for taking the time!

[gerlero]

@gdalle Thanks. Tests are now live and passing so this is ready for review

[gerlero]

Gentle bump @gdalle; can you be a reviewer in this PR?

[gerlero]

@gdalle @devmotion Thanks for the reviews.

Any reason why you didn't add tests for Tracker and RuleConfig backends?

I simply omitted them because I saw that those backends don't have tests for hessian either. I've now tried adding the tests: RuleConfig passes, but Tracker fails (ArgumentError: Tracker backend does not support nested differentiation.). So, I've now added the RuleConfig backend tests, but left Tracker out.

EDIT: the RuleConfig tests fail with Julia nightly though..

[gdalle]

As soon as I understand what's going on with the tags, it's good to merge on my end

[gerlero]

Thanks. Can we get a new release with this feature?

[devmotion]

https://github.com/JuliaRegistries/General/pull/100271