chore: update acp-extension-core to 0.1.4

[zxch3n]

Related issue

Follow-up to #39 and LodyAI/acp-extension-core#7. Companion to LodyAI/Lody#554.

Problem / pressure

The merged goal prompt implementation consumes Core contracts missing from the pinned 0.1.2 package. The Lody workspace override masked that standalone dependency gap. Published Core 0.1.4 contains both goal prompt controls and worktree project identity; 0.1.3 still lacked the goal additions.

Summary

• Pin acp-extension-core to 0.1.4 in package.json.
• Update package-lock.json with the matching version, npm tarball URL, and registry integrity.
• No adapter source changes or release publication.

Visual explanation

Simple change: this is a two-file dependency/lockfile update; the exact version mapping below is sufficient.

Before / after

Before	After
Core pinned to 0.1.2; required goal declarations available only via the workspace override.	Core pinned to published 0.1.4; standalone registry installation provides both goal and worktree contracts.

Test plan

• Inspected the published 0.1.4 tarball for LodyGoalPromptControl, controlActions, promptActions, and LodyWorktreeProject.
• In a separate clone without the workspace override: npm ci --include=dev --ignore-scripts --no-audit --no-fund passed.
• Standalone npm run typecheck passed for both source and examples.
• 52 tests passed across goal commands, prompt lifecycle, goal control transport, native goal events, session fork, and worktree project suites.
• Workspace frozen-lockfile install, Core build, Codex typecheck, and full workspace typecheck passed.
• Formatting, docs, and public-boundary checks passed.
• Full workspace check timed out after two minutes during lint; full-suite success is not claimed. No live Codex E2E or publish/build lifecycle-script validation was performed.

Context handoff

Instructions for reviewing agents

• Review focus: Verify the manifest and lock resolve the same published 0.1.4 artifact.
• Decisions to challenge: Keep the existing exact-version pin policy; do not rely on workspace substitution as evidence of standalone compatibility.
• Plausible failures / evidence gaps: Targeted registry-backed validation passes, but live E2E and release packaging were not exercised; existing host goal-state P1 findings are outside this dependency update.

Authoring context

• User goal / directives: Consume the newly published Core release and submit the Codex dependency update for review.
• Constraints / non-goals: Change only the dependency and lock metadata, without adapter behavior changes or publishing.
• Risk-bearing decisions: Use the verified npm artifact and its integrity instead of an unpublished branch dependency.
• Destructive or irreversible behavior: No data migrations, deletion, or publication.
• Deliberately not done or tested: Full workspace suite, live E2E, and package publishing.
• Unknowns / confidence: Independent npm install, source/examples typechecks, and 52 targeted tests pass; packaging remains a separate release check.