fix(ingest): resolve advert endpoints after node updates

[n30nex]

First or renamed adverts could broadcast missing or old source names because live resolution ran before the node update. Apply payload side effects before resolving inserted observations; optional new-path repeats resolve current identity without reapplying the old advert. Suppressed copies do no live-only endpoint work.

Closes #164. Independent on accepted dev.

Parent db30c9b573357990c41166292e7cf42785c92cc4, head 319df7da5246a71f9693f923d2124bf831eabb82. Focused diff.

Published-head CI and the complete combined native Pi/PostgreSQL suite pass. The Pi runs server a35cba1d / web e1133ab5, retaining every review candidate. Windows/Pi web build/lint and all 940 tests pass. The isolated 3,200-input/504-scope replay preserved expected data/events with zero fixture drops; this is not a universal production losslessness claim. Validation, exact heads and recovery / running source.

Maintainers retain merge/release control; AI-assisted work continues under the contributor's standing authorization.

[n30nex]

@MrAlders0n, please review this focused fix (and have Claude review it if useful). It is independent of the analytics-retention work and can merge separately. Native Pi/PostgreSQL validation passed for 74f16de. The formal review-request API is not available to this contributor account, so I am requesting review here.

[n30nex]

@MrAlders0n, ready for your / Claude's review at 2c5ad5fc. The declared server stack has been refreshed for accepted #170; this PR body now links its current focused diff and dependency. Full native Pi/PostgreSQL tests pass in composed server 99e623c5, with every existing candidate included. Exact-head CI passes.

The preview/source now runs server 99e623c5 / web 3b3abdcc. Migration 041 was tested on a restored copy with unchanged row fingerprints; prior database/binary/web recovery is retained. Server merge order is #167 -> #169 -> #172; #166 remains independent. No upstream merge or production switch was performed.

[n30nex]

@MrAlders0n / Claude: #166 is refreshed onto db30c9b5 at e1c4fa2f70ae6b5efc0054b91433460d07f06ce1 and ready for another look. This was the one source conflict in the coordinated refresh.

The fix now keeps your optional new-path repeat events: inserted observations apply their payload side effects before live endpoint resolution; repeat events resolve the current name without applying an older advert again. Suppressed copies do neither live endpoint lookup nor repeated side effects. First/renamed-advert regressions failed on the upstream ordering and pass here; the added repeat regression checks the latest name and lookup/upsert counts.

Focused diff. Actual-head CI passes. The Pi composition also passes full PostgreSQL tests and a 504-scope replay of 3,200 inputs during blocked route maintenance, with all expected rows/events and zero fixture drops. All pending features are still in the preview; no upstream merges were performed.

[MrAlders0n]

The fix is right, and the bug is still on dev. The lookup on dev still runs before handlePayloadTypeSideEffects. This needs a rebase onto dev (conflict in internal/ingest/packet.go with #178 and #179):

• Keep #178's repeat logic: if inserted || repeat { if inserted { side effects + MarkSent } ; lookup ; event }.
• Drop the "Duplicates emit no packet event" comment. It's no longer true with includeRepeats.
• Drop the InsertObservation override in endpoint_matching_test.go. #179 already sets observationInserted.
• Optional: UpsertNode already returns the node id, so the advert path could skip GetNodeByPubkey.
• An advert repeat test (includeRepeats on, duplicate on a new path) would be a nice addition.

[n30nex]

@MrAlders0n / Claude: addressed at 319df7da.

The #178/#179 integration is retained: inserted observations run side effects/MarkSent before lookup and event; repeats do not reapply old adverts. The outdated duplicate-event comment was already removed, and the redundant InsertObservation override is now gone. The advert-repeat regression passes. I left the optional UpsertNode-ID shortcut separate to avoid changing side-effect return plumbing during this correction.

Published-head CI and the complete combined native Pi/PostgreSQL suite pass. The Pi runs server a35cba1d / web e1133ab5, retaining every review candidate. Windows/Pi web build/lint and all 940 tests pass. The isolated 3,200-input/504-scope replay preserved expected data/events with zero fixture drops; this is not a universal production losslessness claim. Validation, exact heads and recovery / running source.

Please take another look when convenient; no upstream merges were performed.