Only four languages override ucfirst (actually they should probably be overriding uc and lc instead of ucfirst and lcfirst):

languages/classes/LanguageAz.php:       public function ucfirst( $string ) {
languages/classes/LanguageKaa.php:      public function ucfirst( $string ) {
languages/classes/LanguageKk.php:       public function ucfirst( $string ) {
languages/classes/LanguageTr.php:       public function ucfirst( $string ) {

And none of those languages' names begin with i/I. So we could just use a language-neutral (e.g. English) Language object to capitalise the first letter of the sidebar language names.

The problem is not with the names of these languages themselves, but with names of other languages, like "İtaliano".

What I'm saying is, because there are no languages whose names begin with a dotted i (as opposed to the standard Latin-alphabet i), we never need to transform the letter i to its uppercase dotted form in this context, hence for simplicity the English language class could be used for all languages.

This is likely a duplicate of T35643 which is to implement lc() and uc() methods for the Turkish language. I did write an implementation a while ago, but it broke at least the handling of magic words in the parser.