meta(ad): Enzyme custom-rule coverage — operator pullbacks, rule-body constraints, GPU parity

[kylebeggs]

Umbrella for the three open Enzyme-rule issues. The exchange and BC sweeps have declared-transpose rules over raw storage (_exchange_storage! / _bc_storage! + BlockLayout); the operator sweeps do not, one crash is fixed but not root-caused, and none of it is tested on a GPU or a packed field.

Sub-issues

1. feat(ad): linear-operator pullback rules (DESIGN §6) — re-verify against current Enzyme #41 — linear-operator pullback rules (DESIGN §6). The upstream tracker it cited closed in February; first step is to re-run the mixed pointer/isbits MWE on current Enzyme and either unblock or re-point.
2. fix(ad): Enzyme rule on single-grid apply_bc! segfaults on Linux x86_64 #40 — confirm that an allocating rule body is what segfaulted on Linux x86_64. Two changes landed together, so the cause is not isolated; an MWE upstream would turn "rule bodies must not allocate" from folklore into a documented constraint.
3. test(ad): GPU/packed-field AD parity for the exchange rules #43 — packed-field and GPU AD parity for the exchange rules. Decide and document whether GPU reverse mode is supported or refused; today it is untested.

Invariants that bind every sub-issue

• A rule is an exact substitution of a declared transpose, never an approximation, and never fires on a path carrying operator parameters (ScalingOp, Advection with a Field coefficient) — test/autodiff.jl guards dκ.
• Rule arguments may not mix GC pointers with inline floats (Julia 1.12 calling convention); rule bodies may not allocate.
• A local macOS/aarch64 green run is not evidence about Linux x86_64. Push and read the CI matrix.

---

_{🤖 Beep boop — Claude grouped the rules; Kyle is not differentiating himself.}