Add per-operation grouping review guideline for dplyr

[d-morrison]

Summary

• Adds shared/coding/per-operation-grouping.md with guidance to prefer dplyr's .by argument over persistent group_by()/ungroup() pairs when grouping is only needed for one operation.
• References the new fragment from CLAUDE.md alongside the other coding guidelines.

Reference: https://dplyr.tidyverse.org/reference/dplyr_by.html
Example flagged in review: https://github.com/ucdavis/rampp/pull/125/changes/#diff-ef2f6f6745046859ccc877eb26de77f63a99edc851687c9fa053788458f7f731R265-R267

---

Generated by Claude Code

[d-morrison]

Driving this PR to clean — back off until done.

---

Generated by Claude Code

[d-morrison]

ARD summary

Finding	Disposition	Action
Closing sentence stacks three subordinate clauses; i.e. instead of that is	Address	Simplified to two clauses, replaced i.e. with "that is" in shared/coding/per-operation-grouping.md

---

Generated by Claude Code

[d-morrison]

ARD summary — round 2

Finding	Disposition	Action
Opening sentence stacks two subordinate clauses	Address	Split into two sentences; removed "R code that uses dplyr" → "dplyr code"

---

Generated by Claude Code

[claude]

Claude finished @d-morrison's task in 1m 29s —— View job

---

Review

• Read changed files (current state)
• Verify prior finding (opening sentence stacking) addressed
• Scan for new issues (hallucinations, style, correctness)
• Post verdict

---

Prior finding confirmed addressed. The round-2 inline finding — opening sentence at lines 3–5 stacking two subordinate clauses ("When reviewing or writing R code that uses dplyr" and "where the grouping is only needed for one operation") — is fully resolved. Lines 3–5 now match the suggested split exactly: one sentence stating the preference, a second applying the condition.

No new findings. Checked:

• Prose style: Every sentence carries at most one subordinate clause. The closing sentence (lines 17–19) uses an em-dash + "that is" to clarify a single condition; that phrasing was the round-1 suggestion and stands.
• Code examples: Both are correct dplyr 1.1.0+ syntax. summarise(..., .by = group_col) is real per-operation grouping syntax; the group_by() |> summarise() |> ungroup() pattern is accurate.
• Reference URL: https://dplyr.tidyverse.org/reference/dplyr_by.html points to the real dplyr .by documentation.
• CLAUDE.md wiring: The heading and @shared/coding/per-operation-grouping.md fragment reference follow the established pattern for other shared coding guidelines.
• Hallucinations: None detected — all function names, argument names, and the reference URL check out against actual dplyr.

This PR is fully clean.

[d-morrison]

Done — PR is free.

---

Generated by Claude Code