Refactor blob preloading code.

[kmiller68]

While working on #74 @danleh noted that there seemed to be a bunch of dead code in the preloading code. It does seem like much of the logic there could be simplified.

[danleh]

Thanks for filing! Specifically _loadInternal can be removed I think.

[danleh]

Status update: The dead code is removed with #93. The remaining TODO is to merge/simplify the different code paths for caching resources in the shell vs. browser setting, see

JetStream/JetStreamDriver.js

Lines 192 to 195 in 0b92f93

	// TODO: Cleanup / remove / merge. This is only used for caching loads in the
	// non-browser setting. In the browser we use exclusively `loadCache`,
	// `loadBlob`, `doLoadBlob`, `prefetchResourcesForBrowser` etc., see below.
	const fileLoader = (function() {

vs

JetStream/JetStreamDriver.js

Lines 496 to 500 in 0b92f93

	// TODO: Cleanup the browser path of the preloading below and in
	// `prefetchResourcesForBrowser` / `retryPrefetchResourcesForBrowser`.
	const promises = [];
	for (const benchmark of this.benchmarks)
	promises.push(benchmark.prefetchResourcesForBrowser());

@kmiller68: Before refactoring this further, I'd like to understand the existing code a little better. E.g., why is there a retry mechanism in doLoadBlob? This feels like something that should not be necessary with a reliable webserver?

JetStream/JetStreamDriver.js

Lines 840 to 861 in 0b92f93

	async doLoadBlob(resource) {
	let response;
	let tries = 3;
	while (tries--) {
	let hasError = false;
	try {
	response = await fetch(resource, { cache: "no-store" });
	} catch (e) {
	hasError = true;
	}
	if (!hasError && response.ok)
	break;
	if (tries)
	continue;
	throw new Error("Fetch failed");
	}
	const blob = await response.blob();
	const blobData = JetStream.blobDataCache[resource];
	blobData.blob = blob;
	blobData.blobURL = URL.createObjectURL(blob);
	return blobData;
	}

[kmiller68]

@kmiller68: Before refactoring this further, I'd like to understand the existing code a little better. E.g., why is there a retry mechanism in doLoadBlob? This feels like something that should not be necessary with a reliable webserver?

I believe this was added because folks complained the benchmark didn't run on spotty Wifi/cellular connections.