I see many cases in the regression tests that this change fails to improve, for example:
In input/regression/rehearsal-mark-formatters.ly, the new position of the "H" mark is an improvement, but the new position of the "CC" mark isn't. In input/regression/rehearsal-mark-align-priority.ly, the marks over the F clef didn't need to be shifted, and the mark over the G clef didn't benefit from being shifted. https://codereview.appspot.com/553290043/
