Currently, if a nonexistent issue is requested via /browse/FOO-1, a regular HTML page is returned with a 200 OK status response.

For the benefit of web crawlers (the 2.6 URL space is now crawler-friendly), we ought to return a 404 "Not Found" status. The existing HTML page would still returned as the body, so no user-visible difference.

This would also have the advantage of preventing caches caching the error page, if /browse/FOO-1 ever became available.

Similar story with /browse/FOO project URLs. Currently a nonexistent project redirects to the last valid /browse/ URL. So after /browse/FOO (where FOO is valid), /browse/BAR (where BAR is invalid) will return exactly the same page, but with the wrong URL.