L B' F' L' F // EO(5/5)
U2 L2 D L D' // DR, 4c3(5/10)
L B2 L D2 R2 D2 F2 R' // HTR(8/18)
L2 F2 U2 F2 U2 // finish(5/23)

You can get a 20 by doing rzp in 6 to 2e3c and dr in 11, or a 21 from rzp in 6 on (D) D2 U' L' U but it doesn't look like 5 to dr at all.

I had this dr, but delayed solving it too much to find an easy 22:
L B' F' L' F
(R D R' D) // 4c4 in 9, 22 opt

(D' F2 D) // EO(3/3)
(B' L') // DR-4C2E(2/5)
D2 B R F2 R' // DR, 4a2 4e(5/10)
(F2 U2 F) B2 U2 L2 D2 F // HTR(8/18)
U2 B2 D2 R2 F2 // finish(5-1/22)

Had a lot of mediocre drs in 10, this seems to be optimal from everything i had. Missed a 21 because i somehow forgot to switch the 3mover on normal, oops

R' B // EO(2/2)
(L') // DR-4E4C(1/3)
(R2 D U' L2 D' L) // DR, 4b2 0e(6/9)
(D2 R2 [U' R2 B2 R2 B2 U) // HTR(8/17)
(L2 F2 U2] R2 F2) // LS(5/22)

[] = D U2 B2 L2 B2 L2 D R2 B2 D2(+1/23)

Somehow the only 22 from this dr uses 4q, and with all the other drs i didn't get to checking that. Mallard only finds really obscure 20s

Other:
R' B' 
L'
D2 U' F2 B2 L2 D R // 2c3 2e in 10, 22 opt from variations but skull path

R' B' 
(U L')
U R2 D2 L2 U' R // 4b2 4e in 10, 21 opt with 4q

R' F D' R' // EO(4/4)
B' // DR-4E4C(1/5)
F2 U D2 F2 U R2 F' // DR, 4b2 2e(7/12)
(F2 [U2 L2 D' F2 U]) // HTR(6/18)
(F' B L2 F' B') // LS(5-2/21)

[] = U' D' B2 D' L2 D(+1/22)

Pure luck

I had
D (D' R2 D)
R2 F
R' L B2 D2 B // 2c3 in 11 with variations on both sides but bad

But if you do R2 F' you can do D2 F2 L' R F' to dr, which leads to a free 21, but well

(L' R' F D) // EO(4/4)
(R' D2 L D2 R' F) // DR, 2c3 2e(6/10)
F2 U2 R D2 [R F2 R // HTR(7/17)
D2 B2 R2] D2 L2 // S slice(5/22)

[] = R' L2 B2 R U2 {F2 L2(+1/23) // adjust
{} = F B D2 R2 D2 F B'(+1/24)

This dr is 22 opt from switching D2 L, and my double slice is 23 opt but i was messing around with pairs tracing too much this attempt and had little time to find it

(R) B U B R // EO(5/5)
F // DR-4C2E, AR-3C1E, 2 pairs(1/6)
(D' F' D2 F) // DR, 2c4(4/10)
(D) U' F2 U F2 L2 D' // HTR(7/17)
R2 B2 U2 L2 D2 // finish(5-1/21)

Other funny drs:
(B) F D L B
R D' R U2 L // 2c4 in 10, this is literally the first eo i checked

(B) D2 F L B
U2 L F2 U' R // 4a2 2e in 10

This is probably the most fun i've had on a scramble this bad, lmao

D F L B' // EO(4/4)
R F2 R' D // DR, 4c4(4/8)
L' F2 L (L B2 R2 D2 R) // HTR(8/16)
(L2 B2 D2 L2 D2) // S slice(5/21)

([L B2 R2 D2 R L2 B2 D2 L2 D2 L'] F2 L...)

[] = R D2 R2 F2 R' {U2 B2} R2 F2 R'(-1/20)
{} = U D L2 U D'(+3/23)

This is somehow optimal from dr with 4q, bruh

Only other dr:
L U F2 D' R
(F D' B2 U B) // 4a3 in 10, also 23 opt

Mallard only finds 21 from switching this eo and doing jzp to 4c3 in 11 with a somewhat easy finish, i didn't switch cause 22 arm but should've

D' L R' B // EO(4/4)
(U' R2 D' U2 L) // DR, 2c4(5/9)
[U' B2 U' R2 D L2 D'] // HTR(7/16)
L2 B2 U2 B2 R2 B2 L2 // LS(7-1/22)

[] = D' R2 U' F2 D B2 U'(+0/22)

Early backup:
D' L R' B
R
(R2 U2 F2 L' B2 L')
(D2 R2 B2 U F2 D')
(L2 F2 U2 F2 R2 F2) // 23

Kinda stupid for missing the R U L htr when it was available and gives easy finish for once, but i don't thing you can get sub22 from my drs without that. There's also an easy 20 from doing (D) rzp on inverse which i missed

(B' D2 R F) // EO(4/4)
R2 U' F2 U // DR, 4b3(4/8)
(L) D2 B2 L D2 L // HTR(6/14)
U2 L2 U2 F2 & R2 L2 // S slice(6-1/19)

& = F2 U2 B2 L2 F2 D2 B2 R2(+2/21)

Missed 20 from reducing to 4a2 on normal and switching, and there's an 18 from switching 3mover and doing F B rzp, but could've been worse

D (L D') // EO(3/3)
F' B2 R' B2 R // DR, 4b2 2e(5/8)
(R2 L2 D2 L2 B' R2 F') // HTR(7/15)
(D2 L2 F2 R2 F2 D2 R2) // finish(7-1/21)

Missed 20 from an extension i didn't know existed, but not too bad for the first attempt in almost 2 weeks

(U F2 D) // EO(3/3)
(F2 L B2 D2 L F B) // DR, 4b2 2e(7/10)
B2 R2 U2 D2 L' B2 R // HTR(7/17)
(F' B' R2 D2 F B) // finish, found through switching htr and spamming variations(6-4/19)

Nice, ended at 35 minutes, only had this other dr with a bunch of variations because i decided to ditch 5movers after finding these drs at like 12 minutes

(L2 U F2 D)
B2 L R' F' R' B // 4a3 in 10

F2 D F B' U // EO(5/5)
R' // DR-2E3C(1/6)
B' R' F R // DR, 4b2 2e(4/10)
(R2 D2 B2 D2 B' U2 B') // HTR(7/17)
(R2 U2 B2 D2 R' L' B2 R L) // finish(9-2/24)

Other:
F' R' L U2 F'
D
(D' F2 L2 F2 U) // 4b2 4e in 11 variation hell

B2 U' B' F D'
R D2 B2 D2 L2 F // 2c3 6e in 11

(D F B R)
(B)
B2 L2 B2 D' U2 B // 2c4 in 11

Missed a 23 from the 10+2 using a different 7move htr on inverse and a 6move double slice finish instead of 7, skull. The fact that mallard doesn't find a sub23 without rzp in 6 or dr in 12 is so cursed for a 4/6/8 with a decent number of 4movers. A 22 is possible from my 4b2 variation hell, but you have to niss 4a2 2e reduction(which i totally should've done), but then find a very specific htr with a 3move finish, so still kinda hard even if you switch

R2 L' F D // EO(4/4)
U2 R' D2 L D2 B' // DR, 4b2 4e(6/10)
(D2 R') D2 L' F2 R' F2 L' // HTR(8/18)
F2 B2 D2 // lol(3/21)

Spent a ton of time on:
(F L F' U)
L2 U2 L F // 2c4 in 8, but even 22 is barely findable