To be specific: it matches adjacent duplicate comma-delimited items. This limitation determines the need of a new approach to the subject.
I think I've come up with the right one:
Edit: It is not the right one, actually. See below posts.
Action: Replace with regular expression
Field: {comma- or comma+space- delimited list}
Regular expression: (?<=,|\A) ?([^,]*),(?=.*?(?<=,) ?\1(?=,|\z))
Replace matches with:
It's case-insensitive by default (i.e. will also delete items which differ from the others only by casing)
As we may end up with a space at the begininng after removal from comma+space-delimited list I added second action to the group to strip that space.
Here's how it works all together:
From:
aaa,aaa,aaa,bbb,ccc,bbb,ddd,ddd,ddd,ccc,eee,fff,GGG,ggg,iii
To:
aaa,bbb,ddd,ccc,eee,fff,ggg,iii
Yes, I certainly agree with both!