I went to review the W3C page to double check my understanding that the BOMs were only needed for UTF-16 and UTF-32 because UTF-8 did not have big / little endian.
With HTML5 things have changed a bit since I last visited the page.
Older browsers could use …BE and …LE HTTP headers but now the only way is by reading the BOM
tl;dr for UTF-8 yes strip the BOM