Models update

[Knispel2]

General model -> 690 reactions
Emille model -> 680 reactions
Anais model -> 30 reactions
Julien model -> 279 reactions
Emille_30days model -> 87 reactions

[JulienPeloton]

Thanks @Knispel2 -- all model sizes shrink by a factor of 2, is that expected?

[JulienPeloton]

And the CI is failing with large change: https://github.com/astrolabsoftware/fink-science/actions/runs/15810515645/job/44561198337?pr=544

[Knispel2]

Thanks @Knispel2 -- all model sizes shrink by a factor of 2, is that expected?

I started selecting hyperparameters of the model for each expert using optuna (optimizing the metrics of the base forest), reducing the number of trees, so this is expected

[JulienPeloton]

OK thanks!