GNUbg rollout:
Spoiler
The best move was not a choice ;)
It's only by a small margin though. Normally you wouldn't want to break that 5-prime voluntarily but here the resulting position from 12/7 is relatively unsafe, with 3 checkers to clear from the barpoint before we can start bearing off, and no spares on the 6 and 5 points to help us with ugly rolls. If you wonder how 12/7 can not be the best move, look at how fives play after that move.
1. Rollout 7/5 7/4 Eq.: +0.7095
0.8568 0.0728 0.0042 - 0.1432 0.0016 0.0002 CL +0.7470 CF +0.7095
[0.0006 0.0013 0.0011 - 0.0006 0.0017 0.0002 CL 0.0012 CF 0.0014]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 736195168 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
2. Rollout 12/7 Eq.: +0.7011 ( -0.0084)
0.8594 0.0410 0.0021 - 0.1406 0.0049 0.0000 CL +0.7371 CF +0.7011
[0.0007 0.0009 0.0013 - 0.0007 0.0012 0.0000 CL 0.0014 CF 0.0017]
Full cubeful rollout with var.redn.
1296 games, Mersenne Twister dice gen. with seed 736195168 and quasi-random dice
Play: world class 2-ply cubeful prune [world class]
keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
Skip pruning for 1-ply moves.
Cube: 2-ply cubeful prune [world class]
I guess I missed this one, but in 20/20 hindsight I can say that clearing the 7 should have been on the radar. For one, "a point cleared is a point not to be feared". For another, after clearing the 7 point, any 5 (in addition to any 6) forces elmicho off the anchor, usually with lots of hitting options for our side.
Don't know I would have seen itOTB.