full-day workshop: Make the generated devcontainer startup portable and non-blocking

Problem

Step 0 allows a repository name chosen by the learner or workshop leader, but content/full-day/3-codespaces.md hard-codes this startup command:

"postStartCommand": "chmod +x /workspaces/pets-workshop/app/scripts/start-app.sh && /workspaces/pets-workshop/app/scripts/start-app.sh"

It fails when the generated repository is not named pets-workshop. In addition, start-app.sh deliberately waits for both application processes and does not exit, so using it directly as a lifecycle command can leave the Codespaces startup task running indefinitely.

Suggested change

Use the dev container workspace variable or a workspace-relative command, and launch the long-running application through a non-blocking mechanism appropriate for dev containers.

Acceptance criteria

• The command works for any repository name allowed by Step 0.
• The dev container lifecycle command completes after starting the services.
• Flask and Astro logs remain accessible to the learner.
• Both services restart predictably after a Codespace stop/start.
• Service cleanup does not leave duplicate processes.

Comienza con content/full-day/3-codespaces.md y app/scripts/start-app.sh; inspecciona cómo se genera postStartCommand y cómo el script gestiona ambos servicios. Haz pruebas con un nombre de repositorio distinto de pets-workshop y deteniendo e iniciando un Codespace, verificando la finalización del ciclo de vida, los logs visibles de Flask y Astro, reinicios predecibles y la ausencia de procesos duplicados.