load_rss_data() reports beta = z, se = 1 placeholder instead of true per-allele beta/se

[Yining97]

Repo / component: R/file_utils.R::load_rss_data()

Problem.
When a sumstat file provides z but no beta/se, the loader substitutes a placeholder at R/file_utils.R:1432-1438 (beta = z, se = 1). This preserves z numerically but pins se ≡ 1, so top_loci$beta/se end up on an arbitrary scale rather than the true per-allele effect — not comparable to the individual-level fine-mapping path. The true beta/se are recoverable from z plus sample size N and effect-allele frequency (maf), which are already available in the loader, but are not used.