We have a usual dot-product-on-recent state pattern for AR here. However, in the constructor it seems more natural to define the damp_priors $\psi_1 \sim \pi_1,\dots,\psi_p \sim \pi_p$ in the usual order, rather than reverse order (that how it is in the replication note).

I think just putting reverse into the ARStep constructor is the best way here. Noting that Renewal should follow this convention too for style unity across EpiAware. This can be address by addressing #384 (Renewal is only place that uses scan now.)