"ELO" makes sense for Pandemic and other coop games as a measure of how good is the player against the game compared to other players.
I understand the argument, and it is a good argument at first sight, but in practice I disagree, for a couple of reasons:
The first reason is that Elo works well in competitive, 2-players games, with no chance involved (i.e. chess). As you start to discard those assumptions, what Elo measures is less and less clear. In the case of a many-player, cooperative game with chance (e.g. Pandemic), a high Elo is a combination of "this player knows how to play well" plus "this player has been lucky" plus "this player has played with people that play well". What's the weight of the first factor ("how good is the player against the game")? It is impossible to know. Maybe 33%?
One can assume that luck drawing cards and luck in the quality of fellow players will wash out in the long term and the jewels will raise above the sand, but I find that assumption shaky at best. At the end of the day, playing a very good game with players that are "worse" than you will lower your Elo (and it might make them "better" players but it will lower their Elo!).
Which brings us to the second reason why Elo does not make sense for cooperative games, the one that you have stated: Elo in cooperative games can be abused easily by using unsportmanlike behaviours like leaving early when things start to go wrong. This does not happen with Elo in competitive games. The way this has been circumvented in some games in BGA is by never losing Elo (e.g. The Crew).
So we have two kinds of coop games in BGA: those like The Crew were you can never lose Elo and those like Hanabi where you have your games abandoned early by ultra-competitive players because the game is not going perfectly and it will not raise their Elo. In both cases, I think the concept of Elo has been perverted beyond being meaningful or useful at all, and therefore I think it should be removed.