[devin-ai-integration]

What and why

Four packages each carried a private dedupe that spelled sort-then-drop-adjacent-duplicates by hand, which the standard library's slices.Sort/slices.Sorted + slices.Compact already provide. Each site now calls the standard library directly, preserving the result shape the helper had, and the helpers are deleted:

• internal/frontend/repl/compare.go: dedupe(failures) (seen-map then sort, fresh slice) → slices.Compact(slices.Sorted(slices.Values(failures))); failures is still untouched for the later len check.
• tools/referee/xpect/compare.go + resourceset.go: the adjacent-duplicate dedupe after sort.Strings → slices.Sort + slices.Compact on set.missing and set.libraryRoots (both overwritten, so in-place compaction is unobservable; Missing is omitempty, so nil vs empty does not change the JSON). In matches, the at clause's needle is now squeeze([]byte(text)).Text instead of an inline FieldsFunc closure listing the same four bytes as isSpace, so needle and haystack share one definition of whitespace.
• tools/bench/llm/fidelity.go: the in-place dedupe → slices.Sort + slices.Compact at both callers (same backing array, same first-of-run survivor).
• tools/census/grammar/classify.go: the seen-map dedupe (nil on empty, fresh sorted slice) → slices.Compact(slices.Sorted(slices.Values(literals))), which is also nil on empty and leaves the input alone.

Net: 16 insertions, 65 deletions; no behaviour change.

How it was verified

gofmt -l .                       # (nothing)
go build ./... && go build -C tools ./...
go vet ./... && go vet -C tools ./...
python3 scripts/changelog.py check
go test -count=1 ./internal/frontend/repl/                              ok  29.954s
go test -C tools -count=1 ./referee/xpect/ ./bench/llm/ ./census/grammar/
                                                                          ok  0.143s / 170.447s / 0.005s
go test -count=1 ./...                 120 ok; tests/codegen hit Go's 10m default under full-suite load
go test -count=1 -timeout 40m ./tests/codegen                            ok  353.948s (rerun alone)
go test -C tools -count=1 ./...        28 ok
make lint                              ✓ Lint passed (staticcheck + gosec, both modules)

Checklist

• make test and make lint pass locally (go build, go vet, gofmt -l, go test ./... in both modules as above; make lint ✓)
• Tests added or updated for the change — none needed: the existing package suites cover every call site and the helpers were not test-only
• Documentation extended where it already covers the surface (see CONTRIBUTING.md) — not applicable
• Changelog entry added as changes/unreleased/stdlib-dedupe.changed.md, not as an edit to CHANGELOG.md
• baselines regenerated and make docs-counts run if a gate count moved — no count moved
• No internal work-item labels (waves, slices, F4, K5) in the body, docs, or changelog

Link to Devin session: https://nasa-jpl-demo.devinenterprise.com/sessions/6bdfc22599a64e188b0b5e3a1fd0d050
Open in Devin Desktop: https://nasa-jpl-demo.devinenterprise.com/desktop/session/6bdfc22599a64e188b0b5e3a1fd0d050?variant=devin
Requested by: @HuiJun

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no bugs or issues to report.