Henle Urtext edition notates a G sharp, and adds a comment: "Sources give b instead of #, probably an error".
Seems to me G sharp makes a tiny bit more sense (compare the same measure at beat 1 and measure 235, again beat 3 1/2, where you have a-->bflat as the g sharp-->a before..., all minor seconds, can't be sure with Beethoven though...

)
Interesting question.