enhancement: support bare (unquoted) column name inputs via base R NSE

[chris-prener]

Summary

Explore adding support for bare (unquoted) column name arguments in user-facing functions, using base R non-standard evaluation patterns (substitute(), deparse(), match.call() — cf. library(), subset()).

Motivation

Currently, column/variable arguments are passed as character strings. Supporting bare inputs would align with ergonomic R conventions (e.g., library(dplyr) not library("dplyr")) and improve the interactive user experience.

Design constraints

• Must use base R NSE only — no rlang dependency
• Character string inputs must remain supported (non-breaking)
• Pattern: detect whether input is a symbol or character, handle both:

  arg <- substitute(arg)
  if (is.symbol(arg)) arg <- deparse(arg)

• Document clearly which arguments support bare input

Context

• Parent epic: [Epic E] Code & Test Quality Audit ([Epic E] Code & Test Quality Audit #48)
• Depends on: research: evaluate dplyr removal feasibility (108 call sites) #86 (dplyr removal evaluation) — if dplyr stays, this may not be needed since tidy-select patterns are available
• Related: research: evaluate opportunities to reduce package dependencies #37 dependency reduction research; NSE discussion on research: evaluate dplyr removal feasibility (108 call sites) #86

Acceptance criteria

• Identify which user-facing function arguments benefit from bare input support
• Implement base R NSE pattern for those arguments
• Backward-compatible: existing character string usage continues to work
• No new dependencies introduced
• Documentation updated with examples of both quoted and bare usage
• Tests cover both input styles

Notes

This is a follow-on to the dplyr evaluation (#86). If dplyr is retained, tidy-select semantics already provide this. If dplyr is removed, base R NSE is the path forward.