Use of RFC 2119 terminology in MathML 3.0

As far as I can tell the current draft of the MathML 3.0 spec does not 
use RFC 2119 [1] terminology, although in several places (e.g. section 
2.3.1 [2]) specific RFC 2119 terms are used in a way that may be 
consistent with their RFC 2119 definition. In other places, things that 
look like conformance requirements do not have the associated keywords 
(e.g. [3]).

For clarity in determining what is a conformance requirement and what is 
a suggestion, it would be exceedingly useful to phrase MathML 3.0 in 
terms of the RFC 2119 keywords. This will make the spec easier to 
understand and hence to implement and to author against.

[1] http://www.ietf.org/rfc/rfc2119.txt
[2] http://www.w3.org/TR/MathML3/chapter2.html#id.2.3.1
[3] http://www.w3.org/TR/MathML3/chapter3.html#presm.mstyle