Change as little as possible
Decreasing the flaps from 30 to 25 got the a/c an extra 50 m. which avoided hitting the ILS (localiser) antenna array.
There have been suggestions that go-around flaps instead of 25 could have obtained a better result and doubtless simulator trials could establish whether that would have helped or not.
In the abbreviated time available to the crew, any such results were not available.
There is a conflict between obtaining good enough and the best possible. The danger in trying to improve upon good enough in a limited information context is that you might go too far. It's well known that every decrease in flaps increases stall speed, but less well known is that flap retraction can call for lowering the nose to regain the correct speed -- the ground can get in the way of that as the experience with flapped gliders has shown.
Alaska Airlines Flight 261 is instructive in this context. The crew had regained control from loss of stabiliser control and began to vary the flaps. The initial selection of slats and flaps had fortuitously worked well, but retraction put the a/c into a degrading control situation as the already heavily stressed jackscrew gave way.
In BA's case the crew wisely restrained themselves from explorations better confined to a simulator once a good enough result had been assured.