Modernize PP.jl benchmarks

[MSRudolph]

Hi all, thanks for including PauliPropagation.jl in your benchmark. This is a PR showcasing the latest performance possible with PauliPropagation.jl. It requires version 0.8, and uses the VectorPauliSum with the PauliPropagation.Performance sub-module. Also, it performs in-place propagation, which by itself shaves off 4x and 2x (on v0.8.1, but should be similar on v0.7.3) from the baseline (see second timing column). The timings are single-threaded, min of 3 after warm-up, on an AMD Ryzen AI 7 350 CPU. The validation against ppvm matches to 1e-10.

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

qubits	terms	baseline	in-place baseline	this PR	ppvm
8	4701	0.0097	0.0068	0.0012	0.0029
16	19529	0.044	0.037	0.0041	0.012
24	34353	0.24	0.082	0.012	0.025
32	49177	0.33	0.14	0.016	0.040
40	64001	0.74	0.24	0.034	0.058
48	78825	1.24	0.35	0.049	0.079
56	93649	2.33	0.51	0.066	0.106
64	108473	2.75	0.65	0.083	0.140

Heisenberg (same parameters), time in seconds:

qubits	terms	baseline	in-place baseline	this PR	ppvm
6	1022	0.012	0.0068	0.0010	0.0034
8	16324	0.15	0.12	0.015	0.052
10	225353	2.45	1.21	0.20	0.69
12	1174849	11.8	6.03	1.05	4.86
14	2915879	26.4	13.2	1.68	9.10

[github-actions]

👋 Thanks for opening your first pull request against PPVM!

A quick note on contribution terms: by submitting this PR you
agree that your contribution is licensed under the
Apache License 2.0
and that you accept the
PPVM Contributor License Agreement.
Please skim those before a maintainer reviews — opening this PR
counts as your acceptance.

A few things that will speed up review:

• Read CONTRIBUTING.md
  for the workflow, build commands, and style notes.
• Run prek run --all-files locally; CI runs the same checks.
• Use Conventional Commits
  for commit messages.

We'll get to your PR as soon as we can. Thanks for contributing!