DPS only considers time spent in battle and its main application is to maximise balls (hence rewards) from raids that do not require rejoining. When tearing down gyms and dealing with berried Blisseys, higher DPS leads to less time spent in battle, but possibly more time spent healing between battles (which contributes as much to actual time spent attacking the gym).
If TDO was the ideal metric for gym offence, then Lugia and Blissey would be the best attackers (which is obviously not the case).
Gym battles also have timers, so DPS should not be disconsidered (hence, Blissey on offence is out).
I'd say DPS is more important when you want the job done faster, e.g., taking down a gym before a raid starts. If the defenders don't care berrying anyway, then TDO from more bulky attackers, like Metagross, is more important, as it will cost you less healing items.
Which one (DPS or TDO) is more important really depends.
The metric you are looking for is dps*tdo. you need a mon which can get the job done reasonably efficiently but which is extremely potion efficient. The truth of the matter is that having higher dps for faster clear time isn't going to make much of a difference is whether or not a defender has time to GRB their mons.
As a result, mons such as Lugia and Metagross tend to be the best gym generalists, as they get the job done with the highest degree of potion efficiency. However, mons such as blissey (which have next to no dps and thus are prone to timing out) are far less useful.
I would imagine dps*tdo multiplied by some weighting coefficient W based on general typing advantages would be a lot better already. This would require some simulation and may very well include subjective opinion.
Rayquaza should be very high on the list (high dps*tdo*W). W can be 0.95 since dragon does not get resisted often for example.
Machamp would be slightly below as it has lower dps*tdo but similar W. Fighting kills normal but is a burden to face flying types which are very common
Lugia would fall slightly below still (similar dps*tdo to but less relevant coverage)
The problem of not focusing on DPS is that you give more time to the feeder to keep feeding. There's a limit per person of 10 berries every 30 minutes, so DPS-centric gym attack meta helps to make the feeder get to that limit.
Do you really need to revive that much between gym battles? If you have an army of weather boosted Machamp, you can just go with the next one, and after you kick the Blissey out you can heal up. If you're low on potions, revives or Machamp, then you need to consider other trade-off like MM Metagross.
Truth be told, if they wanted to nerf Blissey, they could have just improved the auto-select algorithm, as it takes time to switch battle parties if you have all 5 populated and you can't even order them anyhow.
Remember there are not just 1 account berrying most of the time. In my situation up to 10 to 30 accounts can be feeding grbs in your face, and 10x10 is 100 PER POKEMON. And you add up 100 and 6 pokemon multiaccounters can feed 600 berries at a single pokemon. This is filth. It needs to be fixed, maybe 3 grb UNTIL the defender is knocked out.
Either there's something off about your math or I don't quite understand your situation. Are you saying that you're getting GRBed by accounts at the gym (spoofed and/or physically located) AS WELL AS distance feeders? Because otherwise the distance limit is 60 GRB total per half-hour.
Either way, the above just points out how gym spoofing can remove any sense of fair play in the game.
(Edited for clarity. )
