v1.3.25 CRAN release

jstable 1.3.25

• Fix: svyCreateTableOneJS / svyCreateTableOne2 now display integer counts (no .0 suffix) when n_original = TRUE. Previously, categorical variable frequencies and the n row were formatted as floats (e.g., 12345.0) even after replacing weighted counts with original data.
• Fix: svyCreateTableOne2 now correctly labels the "Overall" column when using addOverall = TRUE with Labels = TRUE. Previously, the "Overall" header was missing, causing column name misalignment.
• Fix: TableSubgroupGLM / TableSubgroupMultiGLM now fully support family = "quasibinomial" for survey-weighted logistic regression. Previously, "quasibinomial" was not mapped to quasibinomial() for svyglm, exp() was not applied to coefficients, and the column was not named "OR".
• Fix: TableSubgroupGLM / TableSubgroupMultiGLM now automatically convert factor outcomes to numeric (0/1) for survey data with binomial/quasibinomial family. Previously, factor outcomes caused "'-' not meaningful for factors" error with svyglm.
• Fix: count_event_by_glm now handles data.table input correctly by converting to data.frame before column subsetting. Previously, data.table objects from survey.design$variables caused "column name 'required_vars' is not found" error.
• Fix: Replaced all deparse(formula) with deparse1(formula) in TableSubgroupGLM, TableSubgroupMultiGLM, TableSubgroupCox, and TableSubgroupMultiCox to suppress "Using formula(x) is deprecated when x is a character vector of length > 1" warnings with long formulas.
• Fix: svyCreateTableOneJS / svyCreateTableOne2 now display actual sample sizes in the n row instead of weighted totals when n_original = TRUE. Previously, the n row showed the sum of survey weights (e.g., 181,174,390) instead of the original sample count (e.g., 23,641). Supports no-strata, single-strata, and compound-strata cases.

jstable 1.3.24

• Fix: .display functions (cox2.display, svycox.display, svyregress.display, geeglm.display, lmer.display) now correctly handle models with interaction terms (e.g., a*b or a:b). This prevents dimension dropping or row matching errors during formatting.
• New: glmshow.display now correctly preserves the offset term (e.g., offset(log(n))) when calculating crude estimates. This ensures statistically accurate univariate results for Poisson and other GLM models by maintaining the same exposure baseline.