325 byte order mark in text/plain

[ralfhandl]

Fixes #325

[Hubert-Heijkers]

These changes are fine but, unfortunately, only apply to text/plain (which in my mind never were an issue as that's US-ASCII anyway, to being explicit is a good clarification).

[Hubert-Heijkers]

I realize the title of this issue being resolved says "use of byte-order mark (BOM) in text/plain responses" but, as I've expressed originally in a comment on the Jira issue, my interest in this topic of BOMs was more related to responses that actually use a Unicode charset, specifically UTF-8, and as such I'd hoped that we could also make explicit that, based on the remarks wrt the Byte-Order Mark in RFC 3629 that OData services MUST NOT (SHOULD NOT if that would break existing rules) send a byte-order mark in UTF-8 responses (read: irrespective of content-type, with UTF-8 charset implied or explicitly specified using charset parameter).

Should I create a separate issue or can we entertain fixing that as part of this issue as well?