BOM is normally used with UTF-16, not with UTF-8 (both of which, along with UTF-32, are encodings of Unicode).
I've worked with lots of minority languages in academic situations, but I've never run into anything that couldn't be encoded in Unicode. There's a procedure for adding characters (or blocks of characters) for characters or character sets that aren't already included. There are fewer and fewer of those. The main requirement is documentation.
On adding new characters to Unicode, as for any commitee there will be rejection and cases where going through the whole process is cumbersome/not worth it.
It's more commonly discussed in the CJK circles, it reminded me of the Wikipedia entry (unsurprisingly with no English equivalent)
More archaic that minority, but one language I had in mind was one using color coded strings and knots representation. There are latin alphabet mappings, so as long as we trust the translation record keeping per se works in Unicode, but if one wanted to keep the exact original writing it would obviously not work out in plain text. I imagined it's not an isolated instance, but I'm also way out of my depth on this one
I've worked with lots of minority languages in academic situations, but I've never run into anything that couldn't be encoded in Unicode. There's a procedure for adding characters (or blocks of characters) for characters or character sets that aren't already included. There are fewer and fewer of those. The main requirement is documentation.