Odach cc support left truncated surv input

[afarnudi]

Adds Support for Three-Column Survival Input in ODACH-CC

This merge request introduces support for three-column survival inputs of the form:

survival::Surv(t_start, t_stop, status)

Key Changes

1. R/pda.R

The construction of ipdata is now determined by the number of columns in the response object.
If three columns are detected, the function extracts (t_start, t_stop, status); otherwise it falls back to the existing two-column (time, status) structure.

if (!is.null(ipdata)) {
    if (ncol(model.response(mf)) == 3) {
        ipdata <- data.table::data.table(
            time_in  = as.numeric(model.response(mf)[, 1]),
            time     = as.numeric(model.response(mf)[, 2]),
            status   = as.numeric(model.response(mf)[, 3]),
            subcohort = ipdata$subcohort,
            model.matrix(formula, mf)[, -1]
        )
    } else if (ncol(model.response(mf)) == 2) {
        ipdata <- data.table::data.table(
            time     = as.numeric(model.response(mf))[1:n],
            status   = as.numeric(model.response(mf))[-c(1:n)],
            subcohort = ipdata$subcohort,
            model.matrix(formula, mf)[, -1]
        )
    }
}

2. R/ODACH-CC.R

In both ODACH_CC.initialize() and ODACH_CC.derive():

• The col_deg variable is updated to reflect whether the response contains two or three columns.
• Adjustments to ipdata related to Prentice weighting are revised accordingly.

3. R/case_cohort.R

prepare_case_cohort() now uses the model control to retrieve the response, ensuring that the correct covariate_list is extracted for both two- and three-column survival inputs.