About that tile-editor. Why can't the drawing field consist of four tiles (2x2)? That way you wouldn't have to guess where pixels meet at the borders.
Also how about an ad hoc preview animation? There were the three congruent frames of the running animation, so it shouldn't be hard for the program to build a preview animation from that; having to play the rom for an actual examination is asking too much.
Rom-hacking has existed for +20 years, how can this be so impractical? Is this really as good as it gets?