Dynamics should not collide with down beams

[Raphael Manfredi]

Here's a little script demonstrating the problem.  Proper dynamics
printing should probably be added to the "regression tests".

-----------------------------------------------------------------------
% The "mp" indication should not collide with the down beam
\version "2.6.3"
\new Staff {
        \clef treble \time 2/4 \key f \major
        \override Staff.NoteCollision
                         #'merge-differently-dotted = ##t
        \relative c' {
                << f4\mp \\ \times 2/3 { f8^( c a) } >>
                << g'4 \\ \times 2/3 { g8^( e c) } >>
                |
        }
}
-----------------------------------------------------------------------

Raphael