Python track: exercise 08's manifest.yml "command" field isn't reliably honored on newer CF CLI versions — needs a Procfile

從 08-deploy-agent-to-cf.md 和練習的 manifest.yml 開始，然後檢查部署說明目前如何指定啟動命令。在 manifest.yml 旁邊加入提到的 Procfile，其中包含文件記載的 web 命令，保留現有的 manifest 設定，並驗證 cf push 會提供啟動命令且應用程式能啟動。

Summary

08-deploy-agent-to-cf.md instructs learners to set the start command via manifest.yml's command: field only:

​```yaml
command: python -m uvicorn server:app --host 0.0.0.0 --port $PORT --workers 1
No Procfile is mentioned anywhere in the exercise.

Effect
On cf push (reproduced on cf.exe version 8.6.1+b5a352a68.2023-02-27), staging fails with:

No start command specified by buildpack or via Procfile.
App will not start unless a command is provided at runtime.
despite manifest.yml's command: field being present and correctly formatted, exactly as the exercise instructs.

This appears to be a known behavior with CF CLI v7+/v8 — the command: manifest field is documented to work, but isn't always reliably honored during staging in practice. A Procfile is the more reliable mechanism. (This may be a Cloud Foundry platform/CLI-version issue rather than something specific to this repo's code — flagging here since it's a real, reproducible failure for anyone following the exercise as written, and the resulting error message gives no indication that a Procfile is the fix.)

Suggested fix
Add a Procfile (no file extension) alongside manifest.yml in the exercise instructions:
​ web: python -m uvicorn server:app --host 0.0.0.0 --port $PORT --workers 1 ​
Safe to include alongside the existing command: field in manifest.yml — redundant, not conflicting, and covers CF CLI versions/configurations where the manifest field alone isn't sufficient.

Environment
Reproduced on a personal SAP AI Core tenant, Python (CrewAI + LiteLLM) track, cf.exe version 8.6.1+b5a352a68.2023-02-27.