CSDP seems to perform decently as a solver: https://plato.asu.edu/ftp/sparse_sdp.html and unlike a lot of others, is actually maintained. Could that be a good contender for a new SDP solver?