Upgrade default priors in CLV `GammaGammaModel`

[ColtAllen]

The default model configuration for GammaGammaModel contains HalfFlat priors, which can create unstable behavior as data drifts in automated pipelines. Weibull priors are a superior option, and also used as defaults for the flagship ParetoNBDModel. The following configuration would work well:

from pymc_extras.prior import Prior

spend_model = clv.GammaGammaModel(
    data=ggm_fit_data,
    model_config= {
        "p": Prior("Weibull", alpha=2, beta=1),
        "q": Prior("Weibull", alpha=2, beta=1),
        "v": Prior("Weibull", alpha=2, beta=10),
    }
)
spend_model.fit()

Note this will also require updating the GGM tutorial notebook.