Remove non-owner continuation, verification and scanner fallbacks after N-2 #908

Problem

Continuation, verification and marker dispatch retain 352 code lines at packages/agents/source/skills/rebase/shared/scripts/continue-rebase.sh:1, packages/agents/source/skills/rebase/shared/scripts/verify-rebase.sh:1 and packages/agents/source/skills/rebase/shared/scripts/marker-scan.sh:1. Scanner logic already has a Go owner, but docs/agent-skills-migration.md:37 explicitly retains its fallback for older binaries. This is the bounded removal after that promise has expired.

Part of #588 and #326.

Acceptance criteria

1. AC1: Given the fallback-removal release, when compatibility evidence is read, then the required scanner and replay commands have completed their two-release warning period.
2. AC2: Given each assigned legacy entry point and a supported binary, when invoked after removal, then its status, streams and effects match the activated Go contract.
3. AC3: Given the generated deployment, when it is inspected, then these three legacy algorithms and their old scanner probe are absent.
4. AC4: Given old completed evidence, when native verification runs after removal, then the existing completion and cleanup contract remains usable.
5. AC5: Given the removed algorithms, when their coverage ledger is inspected, then their still-supported behavior has executing Go coverage.
6. AC6: Given the removal diff, when protected paths are classified, then no owner or constitutional path is changed or deleted.
7. AC7: Given generated packages and deployments, when consistency checks run, then they pass.
8. AC8: Non-functional: retained executable paths contain only fixed forwarding through the shared launcher, with no new shell logic.

Scope and non-goals

Scope: only non-owner continuation, verification and scanner fallbacks, generated copies, fixed forwarding paths and Go integration adjustments.

Non-goal: Bats or Python test retirement, which must land separately in child 11. Non-goal: rebuilding the existing Go scanner, weakening marker exceptions or empty-skip proof, deleting guards or shared libraries, or changing binary resolution. Owner and constitutional shell remains for the single #588 cutover with owner exact-head review.

Approach

Remove the source algorithms and preserve their public calling conventions through #321's launcher. The existing marker probe/fallback branch at packages/agents/source/skills/rebase/shared/scripts/marker-scan.sh:6 disappears only after its own compatibility promise is satisfied. Child 13 and this child use the same recovery format and can land separately after activation. Preserve the known marker candidate/content behavior documented by child 04.

Dependencies

Blocked by: #896, #898, #904, #905, #333, #544, #753.

Prerequisites: #321's launcher and the completed N-2 window. No dependency on child 13 or on the final cutover under #588.

Size

S. Forecast 472 code lines: 352 measured implementation deletions, 30 fixed forwarding lines and 90 Go-integration/deployment adjustments. Prose allowance: 200 lines.

scc could not run in the isolated checkout. Measured baselines count non-blank lines other than full-line # comments, including Python docstrings and embedded code conservatively. Forecasts include implementation, tests, fixtures and registration; they are estimates, not measured future Go. Generated output is excluded only under packages/agents/claude/.apm/skills/, packages/agents/codex/.apm/skills/, .claude/skills/, .agents/skills/ and tokens.json. The source under packages/agents/source/ and copies anywhere else count. The final diff must be at most 800 code lines and 2,000 prose lines, with no exception; split before implementation if either forecast grows beyond its limit.

Verify plan

• AC1: compare command-bearing releases, migration warnings and the actual removal version.
• AC2, AC3, AC8: exercise argument, stream and status compatibility; inspect every generated entry and search for the removed probe/algorithms.
• AC4: feed completed and incomplete legacy evidence to native verification and inspect consumption/refusal.
• AC5: compare removed algorithms with child 11's ledger and Go CI execution evidence; confirm no Bats or Python test retirement is included.
• AC6: classify the complete diff against the pinned owner/constitutional patterns, including removed and renamed paths.
• AC7: regenerate deployments and verify generated consistency and packaging.
• Definition of done: D1 each AC is verified at the merged head, with the commands and results cited in the pull request's Verification. D2 an independent review approves the exact merged head, and every finding has a recorded disposition. D3 required checks pass with no gate loosened. D4 the leak scans the repository specifies are clean, and every post made for this work passes the per-post leak review before it is posted. D5 every commit is signed. D6 the documentation matches the change or the Verification says why there is no impact. D7 every commit has an Assisted-by trailer and the pull request has the provenance:agent-assisted label. D8 the pull request stays within the size limits. D9 follow-ups are filed as issues and listed under Related. D10 the merge closes the issue through a closing reference, the board shows Done, and the final Status: comment names the merge commit.

Risks

Removing the marker fallback early breaks old binaries; narrowing continuation coverage can permit marker commits. Undo: revert removal and regenerate unchanged fallbacks.

Operator authority

None.