SVYLET: Stata module to perform Wald omnibus F-test, Bonferroni, and Compact Letter Display for means/proportions/totals under svy

svylet tests whether a mean, proportion, or total differs across the categories of an over() variable (e.g., survey year) under a complex survey design, and reports the result as a Compact Letter Display (CLD) -- the familiar Tukey/Duncan-style letter grouping from agronomic ANOVA post-hoc tables (categories sharing a letter are not significantly different), adapted here to a design-based Wald test rather than classical ANOVA. It reports a design-based Wald omnibus F-test (with the Korn & Graubard 1990 small-sample adjustment, matching Stata's own default for test/testparm after svy:), all pairwise comparisons with Bonferroni correction, and the resulting letter groups. All three stat() modes share one Mata engine that operates directly on e(b)/e(V)/e(df_r), so the same code works after svy: mean, svy: proportion, and svy: total without depending on margins/testparm chaining correctly after non-regress estimation commands. An optional boot() setting recalibrates the omnibus F-test by prepivoting (Beran 1988), using a bootstrap that resamples the pooled observations across over() categories (Rao, Wu & Yue 1992; Beaumont & Patak 2012) and reassigns group labels at random to enforce the null hypothesis in the resampling itself (Hall & Wilson 1991).