Document longitudinal Tweedie APIMs

[Pascal-Kueng]

Summary

• expand the development generalized-APIM vignette with reproducible longitudinal Tweedie examples for distinguishable and exchangeable dyads
• explain link-scale dyad-occasion covariance, Tweedie dispersion, and generic versus weak identification
• adapt del Rosario and West's random-effects workflow with explicit Gaussian-to-Tweedie qualifications and fixed-effect-first simplification guidance
• add us(), diag(), homdiag(), shared/difference, dispersion, covariance-recovery, and diagnostic examples

Validation

Parsed, rendered, and visually inspected the vignette successfully

The full longitudinal fits currently remain eval = FALSE in the vignette because each takes roughly 2–3 minutes.