Considering U+FEFF is the UTF-16 BOM, not the UTF-8 BOM

Not so. U+FEFF is also the BOM "for UTF-8". It's simply the BOM, period.^[1]

In UTF-16be streams, it's encoded as FE FF.
In UTF-16le streams, it's encoded as FF FE.
In UTF-8 streams, it's encoded as EF BB BF.

These are (some of) the byte sequences looked for, but they're all (encodings of) U+FEFF.

1. It's also the ZWNBSP if found somewhere other than the start of a stream, but that use has been deprecated in favour of U+2060 WORD JOINER.