Slow models for data generation

Top 25 slowest models -- time benchmarks from my local system. See CI report [here](https://github.com/lnccbrown/ssm-simulators/actions/runs/14206895756/job/39814644455?pr=93#step:8:21583).

```
===================== pytest-timer ============================
[tradeoff_conflict_gamma_no_bias-model_conf84]: 213.8598s
[tradeoff_no_bias-model_conf81]: 74.3686s
[tradeoff_weibull_no_bias-model_conf83]: 53.9342s
[ddm_seq2_no_bias-model_conf58]: 53.7685s
[ddm_par2_conflict_gamma_no_bias-model_conf54]: 50.7563s
[ddm_par2_no_bias-model_conf53]: 48.4588s
[ddm_seq2_conflict_gamma_no_bias-model_conf59]: 42.2495s
[ddm_seq2-model_conf57]: 41.4235s
[ddm_mic2_multinoise_no_bias-model_conf72]: 36.4355s
[ddm_par2-model_conf52]: 36.2294s
[tradeoff_angle_no_bias-model_conf82]: 35.0751s
[ddm_par2_weibull_no_bias-model_conf56]: 34.9055s
[ddm_mic2_multinoise_conflict_gamma_no_bias-model_conf73]: 28.8263s
[ddm_par2_angle_no_bias-model_conf55]: 23.9400s
[ddm_seq2_weibull_no_bias-model_conf61]: 23.1836s
[ddm_mic2_leak_conflict_gamma_no_bias-model_conf78]: 20.7561s
[ddm_mic2_leak-model_conf76]: 20.1977s
[ddm_mic2_multinoise_weibull_no_bias-model_conf75]: 19.5574s
[ddm_mic2_ornstein_conflict_gamma_no_bias_no_lowdim_noise-model_conf90]: 18.6490s
[ddm_mic2_ornstein_no_bias-model_conf68]: 18.4518s
[ddm_mic2_leak_no_bias-model_conf77]: 18.3610s
[ddm_mic2_leak_conflict_gamma_no_bias_no_lowdim_noise-model_conf94]: 17.7731s
[ddm_mic2_adj_conflict_gamma_no_bias-model_conf64]: 17.7335s
[ddm_mic2_leak_no_bias_no_lowdim_noise-model_conf91]: 17.5464s
[ddm_mic2_ornstein_no_bias_no_lowdim_noise-model_conf87]: 17.3582s
```

Start with the linked CI report and the listed pytest-timer results to reproduce the slowest model benchmarks. Trace those model names to their implementations and identify a focused optimization scope; done means the targeted data-generation runtimes improve without changing simulator behavior.

Written by the indexing model from the issue text.