My questions are:
Is there any group, which cannot be written as the quotient of a residually finite group by an amenable normal subgroup? Is it possible for large classes of groups?
and
Is there a group, such that every extension by an amenable group is split?