Aurora event You may want to increase the provisioned storage

So I got out of this jam by:

Changed the instance type of the cluster read instance to the next size up (as localstorage is dictated by instance type)
Failed over the write node - now the resized read instance becomes the writer. We can now write to the database.
Create a new read replica at the correct instance type and wait for it to complete creating
Delete the old, local disk full replica.