Ugh, UTF-8 BOM. Many apps can handle UTF-8 but will try to return those bytes as content; maybe ours in 2015 too
I was on the Power Query team when we were improving the encoding sniffing. An app can scan ahead i.e. 64kB, but ultimately the user needs to just say what the encoding is. All the Power Query data import dialogs should let you specify the encoding.
UTF-8 BOM is probably not a good idea for anything other than (maybe) plain text documents. For data, many (although not all) programs should not need to care about character encoding, and if they include something such as UTF-8 BOM then it will become necessary to consider the character encoding even though it shouldn't be necessary.
I was on the Power Query team when we were improving the encoding sniffing. An app can scan ahead i.e. 64kB, but ultimately the user needs to just say what the encoding is. All the Power Query data import dialogs should let you specify the encoding.