#78829Reference index causes duplicate references with different sorting values
Description
If for some reason the ReferenceIndexer fails before completing (which has happened a lot with low max_execution_time / memory_limit values) the DB can contain duplicates which do not get properly removed again. By removing the “sorting” value from the generated hash each relation gets, the problem can be avoided since existing relations get removed.
Assumption: it must not be possible to have two indexed relations between the same two objects, but with different sorting values.