If you're bothered by the idea of negative-area rectangles, there's no need to justify the assumption that b < a, because that's the only way you can assign any meaning to either side of the equation.
The equation isn't about rectangles at all, so I disagree. The rectangles only exist for the purposes of this proof. Any resulting complications are on the proof, not the reader's burden.