Fix secondary app signals in spawned workers

[bm1016bm-svg]

Summary

• preserve each registered app's runtime state when configuring spawned workers
• refresh secondary apps before their servers start
• verify dev-mode built-in HTTP signals remain available on secondary apps

Why

In spawn mode, the primary app received its prepared state through passthru, but secondary apps were re-imported without their per-app mode and config. A secondary app prepared with dev=True could therefore start with production touchups enabled and fail to resolve built-in HTTP lifecycle signals.

Validation

• python -m pytest tests/worker/test_worker_serve.py tests/worker/test_startup.py -q — 19 passed, 1 skipped
• python -m ruff check sanic/mixins/startup.py sanic/worker/serve.py tests/worker/test_worker_serve.py
• python -m ruff format --check sanic/mixins/startup.py sanic/worker/serve.py tests/worker/test_worker_serve.py
• git diff --check

Fixes #3179

[bm1016bm-svg]

CI failures appear unrelated to this change: the worker regression suite passes locally (11/11), while the failing type check is in sanic/pages/error.py and coverage fails in the uvicorn/websockets test environment. Recent main-branch Tests/Coverage runs are also failing. I cannot rerun the workflows without repository admin permission.