setup_readme(): add the has_example argument the guide documents, template example section (supersedes #24)

[larnsce]

The ghedatapublishing guide instructs readers to run setup_readme(has_example = TRUE), but setup_readme() takes no arguments; has_example belongs to setup_website(). The guide documents an API the package does not have.

Tasks:

• Add has_example to setup_readme() (consistent semantics with setup_website's argument), or make one function own the concept and update the guide accordingly
• README template gains an Example section with a commented ggplot2 scaffold, matching the guide's "create a data visualization" step; this absorbs the intent of Templates for plots and tables #24 (plot/table templates), which can then be closed as superseded
• Also fix the fragile first-dataset lookup (list.files("data")[1] yields NA on an empty data dir and only warns after the fact)

Acceptance: the guide's documented call works; a new package gets a README template whose example section compiles once a dataset name is filled in.

Depends on the setup_readme data-loss fix shipping first (v1.0.2). Milestone: v1.1.0 (set manually, see the milestone setup issue).

[larnsce]

Cross-reference: #98 (Rmd to qmd README switch) is now scoped and parked in the backlog milestone, sequenced after this issue. This issue proceeds on the Rmd template for v1.1.0; if #98 lands later it migrates the template as it exists then, example section included. Keep the two in mind together so the template is not rewritten twice.

[larnsce]

Cross-references from the triage in #113 (2026-09-02). #101 (the dead license link in the template) shipped separately in dd99afe through #114, so the template this issue edits already has the working link. #103 (several datasets, dataset named after the package) overlaps the third task here: this issue fixes the crash on an empty data/ directory only, and the multi-dataset loop is the washr skill's job (#115, decision D in #111), with the function gaining at most a small argument. #98 (qmd) still migrates the template as it exists after this issue.

[larnsce]

Done on dev in 7bf433c (2026-09-02), not yet on main. setup_readme() takes has_example (default FALSE, after force so positional calls keep their meaning); with TRUE the README carries an Example section with a commented ggplot2 scaffold for the first data object, which compiles as it is. The empty data/ case now stops before writing instead of warning after a README with NA in it; with several data objects the function says which one the template documents (the loop is #115's). #24 was closed earlier as superseded by this. Tests cover the Example section, the default, and the empty data stop.