[gc_fuzz]: Extend everything we have done so far to array types

[khagankhan]

• types.rs: CompositeType::Array with a reused StructField element. Full array-to-array subtyping (element covariance/invariance, forced element-compatibility in fixup). Cross-kind supertype edges (struct<->array) are cleared. StackType::Array wired into operand satisfaction, eqref upcast, and cast repair.
• ops.rs: a full mirror of the struct op family for arrays, plus array-specific ArrayNew (fixed-length array.new_default), ArrayGet/ArraySet, and ArrayLen. Accesses are null-guarded and use index % (array_length + 1) so out-of-bounds traps are exercised but rare. A same-kind index remap keeps struct ops off array types (and vice versa) now that both share one type-index space.
• mutator.rs: array generation, element mutation, and generic rec-group duplication.
• oracles.rs: take_array + typed take_array_N imports, and Trap::ArrayOutOfBounds added to the accepted-trap allowlist.
• limits.rs: a fuzzable array_length knob (1..=16, default 5).
• config.rs: Config::to_store() could panic in Engine::new() for pooling + GC configs. The GC-heap backing memory rounds its minimum size up to a 64 KiB wasm page, but the harness clamped gc_heap_initial_size to max_memory_size in raw bytes — so a non-page-aligned max_memory_size let the rounded-up minimum overshoot the pool's per-memory limit, and Engine::new (correctly) rejected the config, which to_store().unwrap() turned into a crash. Fixed by clamping the initial size to max_memory_size rounded down to a page.

+cc @fitzgen @eeide

[github-actions]

Subscribe to Label Action

cc @fitzgen

Details

This issue or pull request has been labeled: "fuzzing"

Thus the following users have been cc'd because of the following labels:

• fitzgen: fuzzing

To subscribe or unsubscribe from this label, edit the .github/subscribe-to-label.json configuration file.

Learn more.