fix(runtime): preserve Plan final responses

[Sun-GLiang]

Plan execution completion is a tool boundary, not a conversational terminal. Stop setting the agent-loop stop flag for completed and cancelled Plan executions so the model can consume the durable result and produce a final response.

Keep SubmitPlan as the explicit one-step plan_handoff, and cover completion, cancellation, and submission boundaries with provider-loop regressions.

Fixes #3884

Verification

• npm --workspace @maka/runtime run typecheck
• npm --workspace @maka/runtime run build
• node --test packages/runtime/dist/__tests__/ai-sdk-backend.test.js — 202/202 passed
• npx biome check packages/runtime/src/ai-sdk-backend.ts packages/runtime/src/__tests__/ai-sdk-backend.test.ts
• git diff --check

AI use

Select exactly one:

• No generative tool made a substantive contribution
• Generative tooling made a substantive contribution

Tool(s) and scope: Maka investigated the Runtime Plan terminal path, implemented the loop-boundary change, and authored the regression coverage and PR text. The commit includes a Generated-by: Maka trailer.

Checklist

• Tests cover the change and fail without it
• Lint, format, typecheck and the affected suites pass locally

Does this PR entail a change in behavior?

• Yes — completed or cancelled Plan executions now allow the model's final response; Plan submission still hands off for approval
• No