feat: owner-managed global default hooks on V2 MultiHookRouter

[JhiNResH]

Rewritten on V2 per @ariessa's comment. Closes #28.

Summary

Extends the V2 per-selector, per-hook-data router with owner-managed global default hook lists that apply when a job has no per-job configuration for a given selector.

Resolution per (jobId, selector):

• per-job list non-empty → use per-job (unchanged V2 behavior)
• per-job list empty → fall back to global defaults
• both empty → no-op

All V2 guarantees carry through for resolved lists regardless of source:

• per-hook data dispatch via abi.encode(bytes[])
• de-whitelist resilience (skipped, event emitted)
• selector-completeness validation against IERC8183HookMetadata

Validation at FUND runs unconditionally — even when FUND has no resolved hooks — so a dependent hook declared on one selector without its counterpart globally is caught before value transfers begin.

Admin surface (onlyOwner)

• configureGlobalHooks(selector, hooks) — replace per-selector list
• batchConfigureGlobalHooks(selectors[], hooksPerSelector[][])
• addGlobalHook(selector, hook)
• removeGlobalHook(selector, hook)
• reorderGlobalHooks(selector, hooks)

Views: getGlobalHooks, globalHookCount, resolveHooks.

Events

GlobalHooksConfigured, GlobalHookAdded, GlobalHookRemoved, GlobalHooksReordered.

Tests

23 new tests in test/MultiHookRouterGlobal.t.sol:

• Admin CRUD + per-selector independence
• Revert paths: duplicate, non-owner, invalid selector, not whitelisted, max cap, length mismatch
• Resolution: global-only, per-job override, neither
• Runtime dispatch: global invoked, per-job precedence, no-op, de-whitelist skip
• Cross-selector dependency validation on resolved lists

Ran 23 tests for test/MultiHookRouterGlobal.t.sol:MultiHookRouterGlobalTest
Suite result: ok. 23 passed; 0 failed; 0 skipped

Test plan

• forge build green
• forge test --match-contract MultiHookRouterGlobalTest — 23/23 passing
• Maintainer review

[ariessa]

Hi @JhiNResH,

Please pull current changes from main branch into this PR and resolve conflicts in this branch.

[JhiNResH]

Updated PR #34 in commit 244b2e1.\n\nPulled latest upstream main into the branch and resolved the MultiHookRouter conflict. The branch is now mergeable again, and the diff is back to the router + global-default tests only.\n\nVerification: forge test passes, 23 tests passed / 0 failed.

[ariessa]

Hi @JhiNResH,

Thanks for addressing the requested changes. Appreciate the work.

After thorough review, this PR will be closed rather than merged. The global-defaults model conflicts with the router's core invariant (client-elected hooks, locked at Funded), and the implementation has security regressions on top of that:

• Globals bypass the Funded-state lock — owner can mutate hooks on in-flight jobs.
• removeGlobalHook reintroduces swap-and-pop; fix(MultiHookRouter): preserve hook order in removeHook #41 just fixed this on the per-job path.
• Owner mutation of globals can DoS in-flight jobs via HookDataLengthMismatch.
• _validateSelectorCompleteness runs only at FUND; globals can drift after.