feat: Add reverse/forward counterparts

[rsenne]

PR to add the forward/backward counterparts discussed. I wasn't sure if for a generic backend if there was no reasonable counterpart to return e.g., if one did reverse_counterpart(AutoForwardDiff) if the code should throw an ArgumentError or if it should just return the backend. I opted for the former

Resolves #1025.

[codecov]

Codecov Report

❌ Patch coverage is 96.15385% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.94%. Comparing base (5572e56) to head (1802709).

Files with missing lines	Patch %	Lines
DifferentiationInterface/src/misc/zero_backends.jl	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1036      +/-   ##
==========================================
+ Coverage   98.21%   98.94%   +0.72%     
==========================================
  Files         138      117      -21     
  Lines        8131     5955    -2176     
==========================================
- Hits         7986     5892    -2094     
+ Misses        145       63      -82     

Flag	Coverage Δ
DI	98.94% <96.15%> (-0.03%)	⬇️
DIT	?

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[gdalle]

Thank you for getting this started!
Most test failures are a red herring (test CI runs by forcing the highest possible version of each package, which here leads to incompatibilities between the recent OrderedCollections v2.0 and the not-yet-updated DataFrames)

[rsenne]

I think I addressed everything:

• We now emit a warning (and return the backend unchanged) instead of throwing an error
• Added a testing recipe (test_counterparts) in DIT that I now call in each backend's tests
• Moved the Mooncake counterparts into the extension
• Tested that the Mooncake config is carried over to the counterpart
• Tested that the Enzyme mode attributes are preserved across the flip, plus the <:ReverseMode signature so extra type params won't break DI
• Generic fallbacks now also cover ForwardOrReverseMode and SymbolicMode

Let me know if anything needs further adjustment!

[gdalle]

Almost there!

[rsenne]

okay think i got everything again

[gdalle]

LGTM except

• the warnings to remove
• the test failures for Zygote and ReverseDiff
• the missing coverage

[gdalle]

Thanks, sorry for the long-winded review!

[wsmoses]

this unquestionably uses internal apis and is subject to breakage.

@gdalle / @rsenne instead can I recommend you add a version of this in enzymecore [and then call that version here]. That way it won't get broken if/when we add/change type parameters to the mode

[rsenne]

I can author an upstream PR -- that work for you @gdalle?

[gdalle]

Yeah I have already mentioned it above in the PR, but DI uses these internals at other points cause it definitely needs them. We'll upstream all we can to EnzymeCore, feel free to get the ball rolling @rsenne and I'll follow up