Dead code in PL IR: unused fold_func_param, write-only prev_columns, unreachable display_ids

The nightly survey (2026-04-12) identified several pieces of dead code in the PL IR and related modules. These are structural enough that they may warrant maintainer review before removal.

1. fold_func_param is never called

fold.rs:289 exports fold_func_param but it has zero callers. The default fold_func implementation uses struct spread and only folds body and args, skipping FuncParam::default_value and Func::env. This means all PlFold implementors miss expressions within default parameter values and captured environments.

This could be a latent correctness issue if any fold pass needs to visit those fields, or it could be intentional if those fields are always resolved before folding occurs.

2. Lineage::prev_columns is write-only

lineage.rs:21 defines prev_columns which is populated by Lineage::clear() but never read anywhere. The comment says it is "A hack that allows name retention when applying ExprKind::All { except }", but no code reads it. It is also #[serde(skip)].

3. display_ids parameter is always false

lineage.rs:74 display_lineage accepts display_ids: bool but the only call site always passes false, making the true branch unreachable. Appears to be an unfinished debugging aid.