The new method is not abstract, and the PropertySheet class is not intended
to be extended (@noextend), so it should not break client code for "good"
There is a small chance that someone extended this class (1) and implemented
same method with the similar signature (2), which would break that client.
Practically, I don't think that conditions 1&2 have significant probability
to appear at same time.
Since the bug itself is annoying for clients of the Properties view, I
would like to request the approval from PMC to backport this fix to 4.6.x