They chose M87 because it was the black hole that should have had the largest angular size as viewed from the Earth. It was also a very massive black hole and the long characteristic timescale for variations in the accretion around it mean that many images, taken over several days, could be co-added to produce a clearer final result.
Details:
The linear size of the expected "black hole shadow" is a small multiple of its Schwarzschild radius, which in turn is proportional to its mass. The angular size in radians is then found by dividing this by the distance to the black hole.
Given a telescope with a fixed angular resolution, you get the best image for an object with the largest angular size. When you compare $r_s/d$ for the various candidate black holes, you find that the one in M87 gives the largest value - it is much further away than Andromeda, but much more massive. The black hole shadow in Andromeda is predicted to be about half the angular size of the one in M87.
The only black hole with comparable angular size is the one at the centre of the Milky Way - much less massive, but much closer. An image of this black hole has been published since your question was asked. This image is not as good as the one for M87 because the image is less "stable". The timescale for changes in the accretion environment around a black hole is of order $r_s/c$. This is days for the black hole in M87, but minutes for the black hole in the Milky Way. The longer timescale in M87 makes it easier to co-add images taken over several days. For the Milky Way black hole, special techniques had to be developed to pick out the persistent features from a rapidly varying set of images.
The nearest known stellar black holes are thousands of light years away but a billion times less massive than the M87 black hole. Their angular sizes would be ten thousand times smaller, so they would not be resolved by the EHT. Their images would also be varying on timescales of milli-seconds, which means there is no possibility of producing a stable image.
