... enclosing a regex in 'm()' ...

The
   m open-delimiter pattern close-delimiter
form is what I think of as the "canonical" form of the m// operator, where the delimiters can be a wide variety of characters including {} () <> [] matching braces. The // match form is a special case. Likewise the qr// s/// operators. This alleviates a lot of escape-ology connected with the / character in regexes. See perlop. (Note that q// qq// qx// qw// tr/// y/// and maybe some others also use this delimiter convention.)

What is the trailing xms doing?

I use the /ms modifiers as part of a standard "tail" on all my qr// m// s/// expressions to give the . ^ $ operators a standard | fixed behavior. This eliminates some degrees of freedom in regex behavior and makes them slightly easier to understand. The /x modifier in the standard tail enables the use of whitespace to help clarify a regex. See Modifiers in perlre.

(?: (?! </div) .)+

This has already been covered by GrandFather here. This expression just steps forward grabbing one character after another as long as that character is not a part of whatever matches the (?!...) negative lookahead expression, a closing div tag fragment in this case. A bit slow perhaps, but effective and flexible (update: flexible in that the lookahead expression can be of any complexity). See Lookaround Assertions in perlre; see also perlretut, perlreref and perlrequick.

(</div)?

Optionally capture a literal character sequence if it is present. The capture variable $1 (in this case) will hold the captured sequence if it was present, otherwise $1 will be undefined. See perlre, etc., as above.