Fix docs build: import EnsembleProblem/EnsembleThreads in shapley.md

[ChrisRackauckas-Claude]

Summary

Fixes the failing Documentation CI build (see failing run).

Root cause: The @example shapley block in docs/src/tutorials/shapley.md uses EnsembleProblem and EnsembleThreads but never imports them. Other tutorials (parallelized_gsa.md, juliacon21.md) explicitly import these via using SciMLBase: EnsembleProblem, EnsembleThreads, but shapley.md was missing this import. With SciMLBase v3, these names are no longer transitively brought into scope through OrdinaryDiffEq's re-exports, so the doc build fails with:

UndefVarError: `EnsembleThreads` not defined in `Main.__atexample__named__shapley`

which then cascades into UndefVarError: shapley_effects not defined in the subsequent example blocks that depend on it, terminating the makedocs build with a [:example_block] error.

Fix: Add the missing using SciMLBase: EnsembleProblem, EnsembleThreads import to the first code block of shapley.md, matching the pattern already used elsewhere in the docs.

Test plan

• Instantiated docs/ environment locally and reproduced the exact failure (UndefVarError: EnsembleThreads not defined) with the import missing.
• Applied the fix and ran a scoped smoke test replicating the shapley.md example code (same imports, ODE/NN setup, batched_loss_n_ode, and a gsa(..., Shapley(...)) call with reduced n_perms/n_var/n_outer for speed) — confirms EnsembleProblem/EnsembleThreads resolve correctly and shapley_effects computes successfully end-to-end.
• Did not run the full docs/make.jl (the untouched tutorial as written trains a Neural ODE for 300 iterations and computes Shapley effects with n_outer up to 100, which is prohibitively slow for local verification), but the fix is a one-line addition identical to the working pattern used in two other tutorials in this same repo.

Made with Cursor