Yes, this means I'm going back on myself, errors besides 404 / 410 in POST-side
trigger the failure counter increment. But it also means more flexibility
for the frontend, being able to still retain information about a dead mention,
but communicate it with extra information to tell the frontend that it's dead.

Maybe it would be a good idea to set a "removal" trigger, and use a cron
based task to clean up the dataset, by removing anything past that threshold?