More resilient logic for stripping the origin from page URLs

[derrickreimer]

The previous method for stripping the origin from absolute page URLs produces malformed output when the host string happens to appear multiple times in the URL.

// Old logic
"https://example.com/foo?return_to=https%3A%2F%example.com%2Fbar".split("example.com").pop() === "%2Fbar" 

// New logic
const url = new URL("https://example.com/foo?return_to=https%3A%2F%example.com%2Fbar")
url.pathname + url.search + url.hash === "/foo?return_to=https%3A%2F%example.com%2Fbar"

I had some trouble getting an automated test case to work in response.test.ts (they all seem to failing for me when I try to run vitest 🤔 ), but happy to take another crack at it with some guidance.

[joetannenbaum]

Looks good, thank you!