normalize-for-search. Un-accents and un-umlauts characters in a string. Also preliminary converts the string to lower case. We use it for autocomplete: both for the matched strings -- on the server side, when indexing; and for the strings the user types into a text input in the browser.