Use value_and_pullback_function as a primitive

[sethaxen]

Fixes #34. Breaking change marked with -DEV version number in case it is merged before #35.

[codecov-commenter]

Codecov Report

Merging #36 (3dcb8ce) into master (979fa04) will decrease coverage by 1.56%.
The diff coverage is 81.81%.

@@            Coverage Diff             @@
##           master      #36      +/-   ##
==========================================
- Coverage   80.05%   78.48%   -1.57%     
==========================================
  Files           2        2              
  Lines         396      395       -1     
==========================================
- Hits          317      310       -7     
- Misses         79       85       +6     

Impacted Files	Coverage Δ
src/AbstractDifferentiation.jl	76.47% <81.81%> (-1.75%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 979fa04...3dcb8ce. Read the comment docs.

[mohdibntarek]

This PR needs more thinking. I think when using the Jacobian as a primitive, deriving the value_and_pullback function won't be possible. Only the value and pullback_function is. So it might make sense to define the value_and_pullback_function/value_and_pushforward_function to throw when the Jacobian is used as a primitive. That is users shouldn't expect to use these 2 functions for such a backend or at least the methods won't be define automatically.

[mohdibntarek]

Ah sorry I meant to comment on the other PR. Anyways, I suggest merging this PR and #34 and updating the README once.

[sethaxen]

Ah sorry I meant to comment on the other PR. Anyways, I suggest merging this PR and #34 and updating the README once.

I assume you mean #35? That PR already updates the ReadMe for the redefinition of value_and_pullback_function. Currently definition of a primitive is not documented, so we need no updates in this PR.

[mohdibntarek]

I want to take a stab at this myself if you don't mind. I will open another PR.

[gdalle]

@devmotion should we close this one?

[devmotion]

Probably. We can always re-open if it is preferred over #93.