[WIP] internalization support

[evnchn]

This PR adds internalization support to the NiceGUI documentation, fix #4025 and generally enables multilingual display, despite our documentation created and defined using Python only.

Compared to others libraries which use Markdown, architecturally it is more of a challenge, but we also have a bit more ease just because we are maining one codebase.

Open questions:

• Is this necessary in the age of AI, where people can have AI translation tools?
  • Particularly, can a well-tuned but cheaper LLM (for the batch translation) outperform a non-tuned but more high-quality LLM (used in translation engines, possibly paid)?
• How to break up the file so that we are reading several small files, not one big Python dictionary, so that maintenence is a bit easier and we don't have Merge Conflict Hell in GitHub?
• Tooling such as automatically adding the TRANSLATEID tags, automatically calling LLM for the translation
• Prompt adjustment for the LLM
• Ensure decent enough coverage of the documentation website

Results:

http://127.0.0.1:8080/set_language set the language between:

• en: What we have now
• cn: What was requested in Add Simplified Chinese Documentation #4025
• jp: Generally takes more space because spelling terms in Katakana is less space-efficient than English characters
• de: Concerned over extremely long words messing up the line breaking. Also Zauberzeug is a German company. As a reminder "Kraftfahrzeug-Haftpflichtversicherung" stands for "motor vehicle liability insurance"