scramble: R' U' F D2 R B2 R F2 D2 U2 L' D2 U2 R2 D' L' R' B L' B2 F2 D2 F D' R' U' F

L' U2 R' D B' F' D' B' U' L2

U2 R' U L U' R U F U F'

U L' U' L' B

L' U2 R' D B' // two squares (5/5)

F' D' B' U' L2 // more blocks (5/10)

(B' L2 U' L') // F2L (4/14)

(L U L' U L U' F U' F' L') // fish (10-5/19)

(L U' R' U L' U' R U2) // chips (8-2/25)

The last layer edge case is the one with edges permuted + two opposite edges flipped. One optimal alg for this is

a cancellation of fruruf into Antisune, which also solves the OLLCP case for diag-pi with opposite flip. The mirror/inverse alg solve the same last layer case (at two different angles); one of the angles leads to ab4c and the other leads to ab3c. One of the 3c angles also cancels 5 moves, so of course that's the one I used. Might be the first or second time I'm using this alg in FMC.

The L3C alg at the end (which is just a Niklas + AUF) is also the optimal 3c insertion.

A 31-move solution I found near the end of the attempt (i.e. useless but interesting I guess):

(L' F2 L R F' R F2 D F) // 223 (9/9)

D' B2 D // EO (3/12)

(L' B' L' B2) // F2L-corner; AUF+ab4c2e (4/16)

(B D2 R2 U F U' R2 D B' D) // edges; AUF+ab3c (10-1/25)

(D' R D L D' R' D L2) // finish (8-2/31)

Could also have done a different 10-move EP alg to cancel one more move to get 24 to 5c, but that's even more useless. (The optimal EP alg if you can destroy one slot's corner is 9 moves iirc, but I don't have that one memorised.)