test(core,plugins): remove low-value tests (batch d118) (#162138)

test(core,plugins): remove low-value tests (batch d118) (#162138)

* test(scripts): deslop t0421 tests

* test(node-host): deslop t0420 tests

Remove repeated policy-drift inputs, generic route checks, and replays of
pager, wrapper-depth, and approval-plan owner tests. Consolidate durable
approval fixtures and lifecycle cleanup; infer mock result types instead
of casting partial copies of production contracts. Preserve Vitest routing.

The 50% campaign target is not reached. Retention ledger below names each
remaining case (table rows expanded), using the requested rules:
1 = unique covered execution path; 2 = historical regression; 3 = distinct
security boundary. No unrelated behaviors were combined to reduce counts.

- Native cwd refusal: 2, d88e10a752f, preserve preparation-refusal classification.
- Lost companion response: 3, disabled fallback must never replay locally.
- Already-cancelled command: 1, entry cancellation return.
- Cancelled local completion: 1, post-run cancellation return and signal forwarding.
- Pending Mac cancellation: 1, post-bridge cancellation return without replay/publication.
- Medium-risk auto review: 2, 3b26947669f, ordinary audit-related argv must reach review and execution.
- Auto review without a plan: 3, unbound commands cannot obtain model authority.
- Login-shell startup: 3, startup effects cannot bypass review through a forged plan.
- Reviewer asks: 1, defer-to-human switch branch.
- Reviewer denies: 3, explicit model refusal cannot execute.
- Approved env wrapper argv: 3, preserve approved positional execution semantics locally and over the bridge.
- Transparent versus semantic env wrappers: 3, only transparent carriers may unwrap into an allowlisted executable.
- Nested safe-bin chains: 3, bind rewritten shell payloads to canonical safe-bin paths.
- PowerShell wrappers: 1, non-POSIX transport must bypass POSIX rewriting.
- PATH-token allowlist execution: 3, launch the canonical approved executable rather than a mutable PATH alias.
- Live policy revocation at commit: 3, recheck authority after awaited persistence before real launch.
- Live policy revocation at callback: 3, the launch adapter must recheck immediately before spawning.
- Live policy unchanged at callback: 3, positive control proves authorized real execution succeeds.
- Unbindable auto executable identity: 3, model review cannot authorize an unbound dispatch chain.
- Human executable identity drift at commit: 3, changed executable resolution invalidates the approved command.
- Script drift after commit: 3, recheck mutable operands after the persistence await.
- Runtime script binding at dispatch: 3, reject changed operands and omitted required bindings.
- Relative skill-bin invocation: 3, skill trust cannot authorize an arbitrary relative executable.
- Unsafe environment inputs: 3, block dangerous overrides, argv assignments, and invalid env keys.
- Shell executable env allowlist: 3, filter shell environments even without an inline payload.
- Revoked rule during allow-always commit: 3, do not resurrect revoked authority through grant persistence.
- Unprompted ask tightening during commit: 3, current-policy authority must not survive a newly required prompt.
- Auto-review ask tightening during commit: 3, forwarded model authority remains bound to its prepared policy snapshot.
- Exact-plan forwarded strict inline review: 3, authenticated one-shot authority remains scoped to its bound plan.
- Unavailable screen recording: 3, deny before execution or durable grant mutation.
- Tightened timeout fallback policy: 3, reevaluate fallback authority at commit.
- Fallback without canonical plan: 3, reject unbound fallback provenance.
- Explicit approval without canonical plan: 3, reject unbound explicit authority.
- Delayed approval without policy snapshot: 3, require the prepared policy identity.
- Allowlist revocation after prepare: 3, reject stale authority before committing.
- Mixed fallback and explicit provenance: 3, refuse contradictory authority sources.
- Durable allowlist timeout fallback: 3, positive control for an exact durable grant.
- Revoked durable timeout source: 3, source downgrade removes fallback authority.
- Mac fallback bridge: 3, forward fallback provenance without inventing explicit approval.
- Fallback versus strict inline review: 3, timeout fallback cannot satisfy mandatory review.
- Unknown approval provenance: 3, reject unsupported authority markers.
- Benign awk durable grant: 3, a file-based grant must not reopen inline programs.
- Make inline grant: 3, explicitly approved runtime payloads remain one-shot.
- Windows cmd transport: 3, unwrap transport before deciding shell approval requirements.
- Windows durable trust downgrade: 3, revalidate the exact durable source at commit.
- Durable cwd replacement: 3, directory identity remains bound through execution.
- Safe builtin with redundant exact grant: 1, builtin authorization must survive revocation of an unused grant.
- Socket response loss: 2, adf5c673821, real completed execution must never replay after its reply is lost.

Measurements and validation:
- Executed tests: 90 -> 48, down 46.67%.
- Shard test LOC: 3431 -> 2045, down 40.40%.
- Covered statements: 357 -> 354 (-0.84%).
- Covered branches: 438 -> 435 (-0.68%). Functions: 34 -> 34.
- Only invoke-system-run-plan.ts drops coverage: 3 statements, 3 branches.
  These are removed preparation-only probes; no source file drops >5 statements.
- Both Vitest coverage groups succeeded; no groups were excluded.
- Blacksmith Testbox: both shard files pass (47 infra + 1 unit tests).
  Final file durations: 2.80s infra; 1.95s unit.
- check:changed passes, including core-test types, boundary checks, all
  dead-export scans, and type-aware lint (0 warnings, 0 errors).
- Local oxfmt, plain oxlint, and git diff --check pass.
- No production or shared test-support changes; no bugs or flakes found.
- Proof workflow: https://github.com/openclaw/openclaw/actions/runs/36738903333

* test(ai): deslop t0414 tests

* test(commands): deslop t0423 tests

* test(google): deslop t0416 tests

* test(board): deslop t0409 tests

* test(infra): deslop t0410 tests

* test(gateway): deslop t0429 tests

* test(doctor): deslop t0422 tests

* test(signal): deslop t0426 tests

* test: retain distinct regressions in batch d118