fix: speed up Mooncake by avoiding tuple broadcasting

[gdalle]

using BenchmarkTools, DifferentiationInterface, Mooncake
backend = AutoMooncakeForward();
prep = prepare_derivative(sin, backend, 0.1);
@btime value_and_derivative($sin, $prep, $backend, 1.0);

Before:

julia> @btime value_and_derivative($sin, $prep, $backend, 1.0);
  480.128 ns (14 allocations: 320 bytes)

After:

julia> @btime value_and_derivative($sin, $prep, $backend, 1.0);
  0.916 ns (0 allocations: 0 bytes)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.22%. Comparing base (f8ffd05) to head (eeb4a17).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #853      +/-   ##
==========================================
- Coverage   98.10%   97.22%   -0.89%     
==========================================
  Files         134      131       -3     
  Lines        7988     7966      -22     
==========================================
- Hits         7837     7745      -92     
- Misses        151      221      +70     

Flag	Coverage Δ
DI	97.68% <100.00%> (-1.11%)	⬇️
DIT	96.07% <ø> (-0.33%)	⬇️

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

☔ View full report in Codecov by Sentry.
📢 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.