But it cannot always remain a preprint; at some point the author must decide on the definitive version and publish it; citing a preprint is also problematic if it is in fact a "moving target"; preprints, however, have a decisive advantage: one has published more quickly and can make subsequent corrections.

EDIT: Preprints are actually not a replacement, but essentially loosen the submission deadline and increase the number of (potential) reviewers. But even with this, many disadvantages you criticise still exist; the public reviewers not appointed by the publisher can also be wrong, and the outcry of indignation can be so loud that the author's superiors enforce a withdrawal of the publication.