Add API endpoint to get subtests results

[gsnedders]

Currently, our API doesn't provide any way to get subtest results.

The dashboard does this:

wpt.fyi/webapp/components/test-file-results.js

Lines 271 to 282 in 9c624d8

	// Slice summary file URL to infer the URL path to get single test data.
	resultsURL(testRun, path) {
	path = this.encodeTestPath(path);
	// This is relying on the assumption that result
	// files end with '-summary.json.gz' or '-summary_v2.json.gz'.
	let resultsSuffix = '-summary.json.gz';
	if (!testRun.results_url.includes(resultsSuffix)) {
	resultsSuffix = '-summary_v2.json.gz';
	}
	const resultsBase = testRun.results_url.slice(0, testRun.results_url.lastIndexOf(resultsSuffix));
	return `${resultsBase}${path}`;
	}

It would be much nicer if we just had an endpoint similar to /api/results, maybe even just /api/results/subtest/a/b/c.html, which similarly redirects to relevant file.

[lolaodelola]

This would be incredibly helpful for work happening in ACD. We need to be able to get subtest results so we can provide data on a feature-level e.g. is button supported in chrome. WPT doesn't think about web features in the same way we do so results for a web feature are scattered across folders and files but if we could get sub-test data, it wouldn't matter as we'd be able to query the API and get only the relevant results for the web feature.

[DanielRyanSmith]

If it helps in the interim, you can obtain results at a test level, with subtest details, by looking at the per-test JSON file in cloud storage.

For example, background-fetch/idlharness.https.any.serviceworker.html: https://storage.googleapis.com/wptd/f614ec86ab70acf6c481d4e4d0f92916546c2735/chrome-143.0.7470.0-linux-22.04-0ad3b39440/background-fetch/idlharness.https.any.serviceworker.html

For format is https://storage.googleapis.com/wptd/{sha}/{version}/{test_directory}
This is the same data used to populate views like this: https://wpt.fyi/results/background-fetch/idlharness.https.any.serviceworker.html?label=experimental&label=master&aligned

[lolaodelola]

If it helps in the interim, you can obtain results at a test level, with subtest details, by looking at the per-test JSON file in cloud storage.

Thank you @DanielRyanSmith, this could be an approach we take. I'll do a little more investigation.