I added a couple of constraints to better identify HTML tags still present in titles. I found
- i
- b
- strong
- em
- and various <br>
Current output is at Wikidata:Database_reports/Constraint_violations/P1476
These end up in title and labels. It would be good to strip them before addition, e.g. by your bot at https://www.wikidata.org/w/index.php?title=Q54541874&oldid=687735354