update PauliPropagation.jl benchmark to v0.9.0

[MSRudolph]

Hi all, this is a follow-up to #212 that updates PauliPropagation.jl to v0.9.0. The following timings are on the same AMD Ryzen AI 7 350 CPU, still single-threaded.

One notable change is that the unexported Performance module is now almost fully ported into VectorPauliSum 's default propagation. If you ever do multi-threaded benchmarks, I suggest you use MultiPauliSum{VectorPauliSum} even though the vector-backend could thread.

TFIM (STEPS=10 DT=0.1 J=1 h=1 ATOL=1e-6), time in seconds:

qubits	v0.8.1	v0.9.0
8	0.0012	0.0012
16	0.0039	0.0038
24	0.011	0.0077
32	0.014	0.012
40	0.034	0.021
48	0.049	0.026
56	0.066	0.034
64	0.081	0.042

Heisenberg (same parameters), time in seconds:

qubits	v0.8.1	v0.9.0
6	0.0011	0.00095
8	0.014	0.014
10	0.19	0.13
12	1.02	0.67
14	1.52	1.16