Better SaaS and local architecture

[forrestbao]

Conclusion from tonight's discussion:

1. Two versions of Codepod: local/desktop and remote/server. In the local version, code is stored locally. In the server version, the code is on the server (our SaaS or on-premises private).
2. In the local version, content of a Codepod project is saved as .py files under folder hierarchy. The meta data and the output of pods are stored as comments in respective .py files. In the server version, the content is saved in a DB.
3. The local version of Codepod includes a small daemon and a light app that stays in the system tray. The tray app has a menu option to open the browser and connect to the local Codepod daemon.
4. The server version of Codepod shall have an option to export a Codepod repo to .py files under folder hierarchy. Currently we can export it to a .ipynb Jupyter notebook file.
5. In either local or server versions, users can use local runtime. This means that even in on-premises server deployment, the runtime can run a separate machine.

@lihebi hebi