Improve JS kit performance by precompiling the polyfill

[Angelmmiguel]

Is your feature request related to a problem? Please describe.

Currently, we're appending the source code into the polyfill and compiling it on every request:

wasm-workers-server/kits/javascript/src/main.rs

Lines 85 to 98 in c3a4aa9

	match identify_type(&contents) {
	JSWorkerType::DefaultExport => {
	let _ = context.eval_module("handler.mjs", &contents).unwrap();
	let _ = context
	.eval_module(
	"runtime.mjs",
	&format!("import {{ default as handler }} from 'handler.mjs'; addEventListener('fetch', (e) => {{ e.respondWith(handler.fetch(e.request)) }});"),
	)
	.unwrap();
	}
	_ => {
	context.eval_global("handler.js", &contents).unwrap();
	}
	}

After adding multiple polyfills like TextEncoder and TextDecoder, the JavaScript shims library started to grow. Currently, it's about 1.1Mb (850Kb minimized). This is causing the JS requests to take ~100ms.

Describe the solution you'd like

I would like to precompile the polyfills so we eval the bytecode directly. This should improve the performance as the JS runtime doesn't need to parse and compile the polyfill over and over. To compile source code to bytecode, the Javy toolchain already provides different methods:

• compile_module
• compile_global
• eval_binary

To store the bytecode, we can use the wizer to preinitialize and save it in a global variable.

Describe alternatives you've considered

For now, this is the only alternative I found to improve the runtime performance.

Additional context

No response