Cleaning Tracking Parameters from URLs

Implement clean_url(url) that removes tracking query parameters and normalises the URL:

• drop parameters starting with utm_ and the exact names fbclid, gclid, mc_eid, ref
• keep the remaining parameters (including repeated ones) but sort them by name (stable for equal names)
• lower-case the scheme and host; remove a trailing ? if no parameters remain; drop the #fragment

Use urllib.parse (urlsplit, parse_qsl, urlencode, urlunsplit).