Investigate alternatives for Puppeteer usage

[whimboo]

Follow-up from the discussion on #146.

It would be good to know for what exactly Puppeteer is needed and if it could be replaced with something else that doesn't have a strict requirement for a given browser.

@sidvishnoi maybe you could please explain? I'm happy to help on this issue.

[OrKoN]

@whimboo what's the use case? We could use puppeteer with WebDriver BiDi + Firefox if you need basic JS evaluation to a basic type.

[sidvishnoi]

Puppeteer is used in following cases presently:

1. Getting processed shortname from ReSpec document (after JS evaluation):

   spec-prod/src/prepare-build.ts

   Lines 188 to 193 in bc93d69

   	const shortName: string = await page.evaluate(
   	// @ts-ignore
   	() => window.respecConfig.shortName as string,
   	);
   	console.log("[INFO] shortName:", shortName);
   	return shortName;

2. Extracting bikeshed metadata pre.metadata to get shortname:

   spec-prod/src/prepare-build.ts

   Lines 218 to 237 in bc93d69

   	const text = await readFile(source.path, "utf8");
   	const metadata = await page.evaluate((text: string) => {
   	const doc = new DOMParser().parseFromString(text, "text/html");
   	return doc.querySelector("pre.metadata")?.textContent || null;
   	}, text);
   	if (!metadata) {
   	throw new Error("Failed to read metadata");
   	}
   	const config = getConfigOverride(metadata)!;
   	// Bikeshed allows metadata keys to be in any case: Shortname/shortName etc.
   	const shortnameKey = Object.keys(config).find(
   	key => key.toLowerCase() === "shortname",
   	);
   	if (!shortnameKey || !config[shortnameKey]) {
   	throw new Error("No `shortname` found in metadata");
   	}
   	const shortName = config[shortnameKey];
   	console.log("[INFO] shortName:", shortName);
   	return shortName;

3. Some query selector processing to extract metadata from published pages:

   spec-prod/src/prepare-build.ts

   Lines 263 to 266 in bc93d69

   	const thisURI = await page.$$eval("body div.head dl dt", elems => {
   	const thisVersion = (elems as HTMLElement[]).find(el =>
   	/this (?:published )?version/i.test(el.textContent!.trim()),
   	);

4. Full page content parsing, JS/CSS evaluation with subresources to extract all directly and indirectly referenced assets from a page that need to be copied for deployment:

   spec-prod/src/build.ts

   Line 169 in bc93d69

   async function findAssetsToCopy(source: Input["source"]) {