Presumably the same way it's justified running five car trains up and down valuable ECML paths. Personally I'm not wholly sold but hey ho.
The test isn't "does not abstract any revenue" but instead "is not primarily abstractive". Which means that an OAO is expected not to just take revenue away from a franchised operator but to produce its own as well. So an OAO which just operated between Preston and London would be primarily abstractive. It introduces some competition and may grow the market a bit but it will primarily be taking revenue away from the franchised operator. However one which operates between London and Blackpool will still be abstracting revenue from the franchised operator but will also be growing its own revenue on the new section by improving links between stations either not currently served by the franchised operator or not well served by the franchised operator. Therefore it's not primarily abstractive.
Now, in the case of London to Blackpool, there is a strong argument that the ORR have had to bend their tools almost to breaking point to get the Alliance application to pass the test (I recall reading the original decision letter and it seemed like some mental gymnastics were being employed) which is perhaps why, on first inspection, it does seem odd that it's been authorised.
Yes, by people who don't understand how railways work. A railway is a natural monopoly and if you want an integrated network that includes services which are not profitable then it becomes very hard to have effective competition between operators without massive physical duplication of infrastructure (see the pre-grouping railway of circa 1820 - 1920). If the goal had been effective competition between private operators then we should never have bothered with franchises and the paths should simply be sold to the highest bidder. That would give you the competition that was promised but I rather suspect the outcomes would not be desirable to the majority of passengers.
It's a funny visual though isn't it?