Sch: Fix races from NG Sch work, and in proc_states

[jpsamaroo]

Should fix hangs in array/allocation suites.

Written by Claude Sonnet

[github-actions]

Dagger benchmarks: dirty vs master

	master	dirty	master / dirty
array/dagger/N=1024 (block 512)/add (X + X)	10.4 ± 0.74 ms	9.95 ± 1.8 ms	1.04 ± 0.2
array/dagger/N=1024 (block 512)/alloc (rand)	3.82 ± 0.23 ms	4.48 ± 0.68 ms	0.853 ± 0.14
array/dagger/N=1024 (block 512)/broadcast (X .+ 1)	7.07 ± 0.093 ms	7.36 ± 0.7 ms	0.961 ± 0.092
array/dagger/N=1024 (block 512)/map (sin.(X))	12 ± 0.83 ms	11.4 ± 1.2 ms	1.05 ± 0.13
array/dagger/N=1024 (block 512)/norm	3.87 ± 0.43 ms	3.82 ± 0.57 ms	1.01 ± 0.19
array/dagger/N=1024 (block 512)/reduce (sum)	7.08 ± 0.74 ms	7.85 ± 0.53 ms	0.902 ± 0.11
array/dagger/N=1024 (block 512)/transpose (permutedims)	9.4 ± 2.7 ms	9.35 ± 3.1 ms	1.01 ± 0.44
array/dagger/N=256 (block 256)/add (X + X)	3.51 ± 0.39 ms	3.42 ± 0.056 ms	1.03 ± 0.12
array/dagger/N=256 (block 256)/alloc (rand)	1.41 ± 0.13 ms	1.17 ± 0.27 ms	1.2 ± 0.3
array/dagger/N=256 (block 256)/broadcast (X .+ 1)	2.9 ± 0.088 ms	3.56 ± 0.068 ms	0.814 ± 0.029
array/dagger/N=256 (block 256)/map (sin.(X))	3.54 ± 0.033 ms	4.13 ± 0.37 ms	0.858 ± 0.078
array/dagger/N=256 (block 256)/norm	1.98 ± 0.098 ms	1.99 ± 0.17 ms	0.995 ± 0.097
array/dagger/N=256 (block 256)/reduce (sum)	3.54 ± 0.68 ms	4.05 ± 0.19 ms	0.872 ± 0.17
array/dagger/N=256 (block 256)/transpose (permutedims)	2.53 ± 0.13 ms	2.59 ± 0.068 ms	0.975 ± 0.056
linalg/dagger/N=1024 (block 512)/cholesky	0.0346 ± 0.0055 s	0.0321 ± 0.0019 s	1.08 ± 0.18
linalg/dagger/N=1024 (block 512)/lu	7.61 s	7.54 s	1.01
linalg/dagger/N=1024 (block 512)/matmul (A*A)	0.0619 ± 0.0058 s	0.0555 ± 0.0037 s	1.12 ± 0.13
linalg/dagger/N=1024 (block 512)/matvec (A*x)	10.9 ± 2.4 ms	9.51 ± 0.57 ms	1.14 ± 0.26
linalg/dagger/N=1024 (block 512)/qr	0.379 ± 0.036 s	0.398 ± 0.0025 s	0.952 ± 0.09
linalg/dagger/N=1024 (block 512)/solve (A\b via lu)	10.6 s	10.9 s	0.973
linalg/dagger/N=1024 (block 512)/syrk (A'*A)	0.0428 ± 0.0056 s	0.044 ± 0.0016 s	0.973 ± 0.13
linalg/dagger/N=256 (block 256)/cholesky	7.49 ± 2.1 ms	7.07 ± 1.2 ms	1.06 ± 0.35
linalg/dagger/N=256 (block 256)/lu	1.32 ± 0.014 s	1.34 ± 0.027 s	0.988 ± 0.023
linalg/dagger/N=256 (block 256)/matmul (A*A)	4.29 ± 0.47 ms	4.24 ± 0.47 ms	1.01 ± 0.16
linalg/dagger/N=256 (block 256)/matvec (A*x)	3.54 ± 0.46 ms	3.63 ± 0.27 ms	0.974 ± 0.15
linalg/dagger/N=256 (block 256)/qr	10.6 ± 0.5 ms	11.9 ± 1.4 ms	0.893 ± 0.11
linalg/dagger/N=256 (block 256)/solve (A\b via lu)	2.17 ± 0.011 s	2.22 ± 0.013 s	0.974 ± 0.0076
linalg/dagger/N=256 (block 256)/syrk (A'*A)	7.15 ± 0.93 ms	6.61 ± 1.4 ms	1.08 ± 0.26
stencil/dagger/N=1024 (block 512)/alloc (neighbors Wrap)	0.115 ± 0.021 s	0.106 ± 0.01 s	1.09 ± 0.22
stencil/dagger/N=1024 (block 512)/assign (const)	3.96 ± 0.21 ms	4.09 ± 0.24 ms	0.97 ± 0.076
stencil/dagger/N=1024 (block 512)/multi-expr	10.8 ± 0.26 ms	10.1 ± 0.2 ms	1.07 ± 0.033
stencil/dagger/N=1024 (block 512)/neighbors (Clamp)	0.0564 ± 0.0032 s	0.0484 ± 0.0027 s	1.16 ± 0.092
stencil/dagger/N=1024 (block 512)/neighbors (Pad)	0.0912 ± 0.0091 s	0.0554 ± 0.0037 s	1.65 ± 0.2
stencil/dagger/N=1024 (block 512)/neighbors (Reflect)	0.0564 ± 0.0068 s	0.0548 ± 0.0027 s	1.03 ± 0.13
stencil/dagger/N=1024 (block 512)/neighbors (Wrap)	0.0573 ± 0.0032 s	0.0553 ± 0.0023 s	1.04 ± 0.072
stencil/dagger/N=1024 (block 512)/update (+)	6.34 ± 0.041 ms	6.39 ± 0.2 ms	0.991 ± 0.032
stencil/dagger/N=256 (block 256)/alloc (neighbors Wrap)	16.5 ± 0.93 ms	16.5 ± 0.41 ms	1 ± 0.062
stencil/dagger/N=256 (block 256)/assign (const)	2.49 ± 0.31 ms	2.58 ± 0.14 ms	0.967 ± 0.13
stencil/dagger/N=256 (block 256)/multi-expr	4.62 ± 0.4 ms	4.62 ± 0.28 ms	0.998 ± 0.11
stencil/dagger/N=256 (block 256)/neighbors (Clamp)	13.9 ± 0.28 ms	14 ± 0.46 ms	0.996 ± 0.038
stencil/dagger/N=256 (block 256)/neighbors (Pad)	15.6 ± 0.51 ms	14.2 ± 0.39 ms	1.09 ± 0.046
stencil/dagger/N=256 (block 256)/neighbors (Reflect)	14.1 ± 1.6 ms	13.8 ± 0.076 ms	1.02 ± 0.11
stencil/dagger/N=256 (block 256)/neighbors (Wrap)	13.4 ± 0.18 ms	13.9 ± 0.9 ms	0.964 ± 0.064
stencil/dagger/N=256 (block 256)/update (+)	3.22 ± 0.72 ms	3.66 ± 0.22 ms	0.881 ± 0.2
time_to_load	0.989 ± 0.017 s	1 ± 0.01 s	0.984 ± 0.02

Plots

No regressions beyond 25.0% 🎉

Improvements (> 25.0% faster)

• stencil/dagger/N=1024 (block 512)/neighbors (Pad): -39.3%

Full results and plots (download the benchmark-results artifact).