wp_redirect sends incorrect url when it contains non escaped unicode characters

[louyx]

TL;DR:
When using wp_redirect to a url that contains non-ascii letters, those letters are stripped out from the HTTP Location header.

I have created a small fix for this issue by escaping every non-ascii character:

$redirect_url = preg_replace_callback('/([^\x00-\x7F])/i', function($matches) {
  return urlencode($matches[1]);
}, $redirect_url);

More Details:
Some plugins redirect to some urls containing unicode characters without escaping them first. Escape functions esc_url and esc_url_raw do not solve this either. I think solving this at core level is the best thing to do, especially that it doesn't affect the current behavior of the function.

Here's an example (in Arabic, which is RTL):

wp_redirect( 'http://example.com/tag/تجربة-1/' );
//redirects to http://example.com/tag/-1/

[pento]

In 31587:

When sanitizing a URL to redirect to, UTF-8 characters can be URL encoded, instead of being removed.

While RFC 3986 does not specify which character sets are allowed in URIs, Section 2.5 states that octects matching UTF-8 character encoding should be percent-encoded, then unreserved octets outside of the UTF-8 range should be percent-encoded. As browsers tend to only implement support for UTF-8 in URLs, this change only implements the UTF-8 encoding part. We may revisit the second part if it becomes an issue.

Fixes #31486