Give the SRI weak-order study enough trajectories for its own tolerance

[ChrisRackauckas-Claude]

Please ignore until reviewed by @ChrisRackauckas.

Fixes #4061.

The problem

iip_weak.jl's SRI study asserts abs(sim.𝒪est[:weak_final] - 2) < 0.5 from 2e4 trajectories, and fails:

Test Failed at lib/StochasticDiffEq/test/weak_convergence/iip_weak.jl:186
  Expression: abs(sim.𝒪est[:weak_final] - 2) < 0.5
   Evaluated: 0.5533865655544976 < 0.5

It is deterministic — it reproduces to the last digit across runs — and it is not new. The group simply never got this far: the file was OOM-killed at the SROCK2 call on line 71, 115 lines earlier, until #4060 landed. Everything after line 186 had never executed.

Why more trajectories rather than a wider tolerance

Sampling the estimator over 12 independent seeds:

trajectories	𝒪 range	mean	sd	max |𝒪−2|	fails < 0.5
2e4 (current)	[1.822, 2.277]	2.006	0.172	0.277	0/12
1e5	[1.892, 2.214]	2.008	0.100	0.214	0/12
4e5	[1.910, 2.078]	1.972	0.049	0.090	0/12

The mean sits on 2 at every count and sd falls as 1/sqrt(n) to within noise — 1e5 → 4e5 is a 4× increase in samples and sd goes 0.100 → 0.050 predicted, 0.049 observed. So the estimator is unbiased and the sole problem is spread: 2e4 gives it a standard deviation of 0.172 against a 0.5 tolerance, and the file's own RNG state (seeded once at the top, then advanced by the fifteen studies before it) lands about 3σ out.

At 4e5 the worst of 12 seeds is 0.09 from 2, more than five times inside the existing bound, so the assertion still has the resolution to catch a genuine order regression. Widening the tolerance to admit a 0.553 estimate would not — it would remove exactly the signal the test exists to provide. The cost is ~4 minutes for that one study (16.3 s → 254.1 s here).

Verification

Whole group through the normal harness, in a 16 GiB cgroup (systemd-run --scope -p MemoryMax=16G -p MemorySwapMax=0) at JULIA_NUM_THREADS=2, matching the self-hosted pool and its num-threads: auto default:

GROUP=IIPWeakConvergence julia --project=. -e 'using Pkg; Pkg.test()'

     Testing StochasticDiffEq tests passed
        Elapsed (wall clock) time (h:mm:ss or m:ss): 17:00.35
        Maximum resident set size (kbytes): 1623164

The log reaches SRI and SRIW1, i.e. this is the first end-to-end run of the file.