Add JS serialization

[Pfeil]

• Provide basic JS generator and boilerplate with roughly correct code, sharing most logic with the other generators
• JS generator needs flag: traces yes/no
• Translate TS boilerplate to JS at build time
• Add ability to fetch and add boilerplate to JS generator
• JS Zip Download contains one file including boilerplate
• Proper sandbox execution trigger in UI
• Make sure the Sandbox retrieves the code including boilerplate
• Check if the syntax is fully valid
• make jsonpath functionality available in boilerplate
• jsonpointer functionality available in boilerplate
• Make it run through completely / successfully
• Consider a way to add jsonpointer.js and jsonpath.js ESM modules via package.json instead of having them in the repo directly
• fix: Download button with JS packages code missing the boilerplate

[craquet]

I was able to find out already that the UI hanging was not your fault at all, everything looks good in that regard. Its actually the creation of a new Context in quickjs that hangs the UI for about 5 seconds every time. If we dont use the context, then there is no hang. I will look into it more

[craquet]

The research concludes:

• Importing the async QuickJS runtime takes the 5 seconds that hang the UI
• It makes no sense for us to use the async runtime, as it does not have to interact with async host code
• WASM code runs in the main thread, always blocking the UI no matter if we use the async runtime or not

Solution: Use the normal QuickJS WASM runtime in a Web Worker (which runs in a separate thread)

Luckily, I have plenty of experience with web workers from NovaCrate, so I'll try to quickly take care of it

[Pfeil]

fixed. On my way to finalize things.

[Pfeil]

Notes:

• The button to run the code is still included. I think it might be useful for development as we do not yet have a gui yet.
• It will always run the JS code. I assume we'll remove this button anyway as soon as we have a GUI, and it does not make any sense for python.

[craquet]

Did not do a thorough review yet, but I noticed that the trace seems to break at least if-else blocks and functions with returns. We have to check if we can find a stable ignore list to ignore these edge cases or if we should take a different approach to tracing.

Maybe we could also leave the tracing out alltogether for this PR, as it is not central to the purpose of this PR and we have to get done until the conference :D

[Pfeil]

• trace: Will give it another try. The trace functionality can be deactivated with a flag internally if we want to.

[Pfeil]

I think I improved the blacklisting in tracing properly now. But I also disabled it (see commit b16bb90) for now.

[craquet]

I only noticed that the otherwise blocks conditionals were flipped (returning otherwise even if either is truthy). Fixed that already. Everything else looks good