Improve youtube and vimeo link-identifying regexps

[gadenbuie]

The current regexps for identifying YouTube and Vimeo links are designed to find links that are known candidates. For example, the YT regexp is good for extracting the embed ID from a known probable YT link. They don't however, include some baseline checks such as checking that the link is an https? link. It would be unusual, but it's still possible that something not intended to be used as a YT link (a base64-encoded image, for example) could be picked out as a YT link.

learnr/learnr-js/tutorial/tutorial.js

Lines 404 to 405 in 1c01ac2

	const youtubeRegex = /^.*(youtu.be\/|v\/|u\/\w\/|embed\/|watch\?v=|&v=)([^#&?]*).*/
	const vimeoRegex = /(?:vimeo)\.com.*(?:videos|video|channels|)\/([\d]+)/i

Our docs state that the whole link should be used, so I think it'd be okay for us to add in ^http to the regexp. Alternatively, we could find a more robust regular expression.

[gadenbuie]

Or we could check that src doesn't start with data: