The characters are selected because they possess specific characteristics that make them ideal to handle various cases.
p possesses maximum "descent" for a character (the bit below the baseline on a font).
É not only meets the traditional "ascent" but exceeds it since diacritics normally extend past the traditional ascent on a capitalized letter.
These aren't the only characters that possess these values (although, I assume some selected code points are unique), but just ones that do.
Also, why use these weird characters (X, p, É, Chinese characters) for the test glyphs? What's wrong with simple ASCII?