v1.1.0: cross-round comparison view (#29) + list pagination (#33)

[mondial7]

The v1.1.0 milestone: the best next enhancement plus the requested performance work.

#29 — Cross-round comparison (growth over time)

The most on-purpose feature for a 360 tool. A subject can now see how their peer competency scores trend across their shared rounds:

• A new server-rendered multi-series line chart (view.LineSVG, no JS lib — consistent with the existing radar/donut SVGs).
• A round-by-round table of per-competency averages and a summary timeline.
• Linked from My Feedback once ≥2 rounds are shared; a friendly empty state below that.

Read-only over consolidations the subject already has access to — no manager-only data, no new privacy surface.

#33 — Pagination

The Rounds (admin), Audit log, and Users lists are paginated (25/page) with Prev/Next controls, via a shared pageNav helper and paged repo methods (FindPaged, fetch pageSize+1 for has-next).

Post-review hardening

A code review over the diff surfaced no runtime bugs but flagged real quality items, addressed here:

• Deterministic paging: paged feedback_rounds/audit_logs queries now carry a unique id tiebreaker so page boundaries can't skip/duplicate rows when created_at collides. A gateway test (5 rows sharing a timestamp, paged in 2s) proves every row appears exactly once.
• Consistency: pageNav is a value type across all three handlers (was a mix of pointer/value).

Verification

go build, go vet, gofmt, go test ./... (incl. Postgres gateway tests) all green. gosec 0 / govulncheck 0. Release Docker image builds and serves the new routes. New behavior covered by unit + handler + gateway tests. CHANGELOG stamped [1.1.0], sonar bumped to 1.1.0.

Known follow-ups (correct today, optimize later)

• The list pages still build a full user index for name resolution — fine now, bound it with a batch FindByIDs as the user base grows.
• The audit filtered view caps at the most recent 200 (unfiltered is fully paged); DB-side filtering would let it page too.