I didn't see SuperTux as a particularly great starting point given what I
wanted to do. Still don't. Actually, if I were to use an existing engine, a
re-implementation of Super Mario Bros X called "PGE" would have been the most
likely candidate:
http://engine.wohlnet.ru/
Incidentally, I think that's a great project. If you're a C++ programmer and
interested in doing something useful for libre software gaming, you might
want to contribute to that. SMBX is not only a quite comprehensive engine,
there have also been tons of levels made for it. PGE getting into a fully
working state would be a huge contribution to libre software gaming as a
result.
But although PGE would have been a pretty good candidate as a starting point,
even that wouldn't have necessarily been easier, and I wanted to test and
grow the SGE Game Engine further, anyway. I think this decision was
absolutely worth it. It has resulted in countless bugs being fixed, and
improvements being made, to the SGE.