Improve detection of valid URLs

Currently the utility function to detect a valid URL is using the GUri parsing API. But apparently this just matches on matching URL scheme (and does not take into account that http and https should start the host part with //. A better approach would probably be to just string match on "http://" and "https://", i.e. something like: url.startsWith('http://') || url.startsWith('https://'). Should also add a test case in test/utilsTest.js:validWebsiteTest ensuring an invalid URL such as "http:invalid.com" is detected as not valid.