Reduce memory footprint with Dockerless app

[onyb]

Currently, ecPoint-Calibrate runs as a set of 3 Docker containers:

• Electron app
• Python backend
• Logging infrastructure

Each of the above Docker containers run full-blown operating systems, which is doesn't leave a lot of memory for computations in resource-constrained environments. Furthermore, what's displayed on the screen is actually a raw stream of the display inside the Docker container, which is often the cause for sluggish user experience.

This issue proposes a lean runtime for ecPoint-Calibrate, without using Docker. This is also a first step to running ecPoint-Calibrate on MacOS and Windows.

Checklist:

• [x] ~Use socket.io to stream logs to the frontend.~
• [x] ~Decommission the current logging infrastructure.~
• [x] ~Investigate PyInstaller, to produce a zero-dependency binary for the Python backend.~
• [x] Use Electron Builder to package the who software into a single AppImage.
• [ ] Implement auto-update.

UPDATE: items that are struck-through have been either evaluated to be unnecessary, or better alternatives were found for them.

[onyb]

Dockerless app shipped in v0.9.0, and continuous deployment implemented with GitHub Actions.

Implementation of auto-update feature (i.e., not having to manually download new versions) is kept as a topic for future.