fix confusion around `p_outlier`

[digicosmos86]

Right now, in HSSM, when p_outliers is specified to 0 or None, it is not added to list_params, so relevant computation could be avoided altogether. However, this causes confusion when people create custom models with make_hssm_rv and make_distribution, because they have to decide whether to include p_outliers in their custom RandomVariable and Distribution classes depending on whether they use p_outliers in HSSM. It's probably a good idea to always include p_outlier in HSSM, so there is consistency across the board