fix(go-core): align thread participants

[casey-brooks]

Summary

• Fixes go-core TestNoDuplicateWorkloads: thread create invalid participants (initiator included) #145 by keeping TestNoDuplicateWorkloads cleanup robust while allowing E2E to exercise the fixed threads service behavior from fix: dedupe CreateThread initiator participants threads#36.
• Reverted the createThread helper compatibility/filtering workaround so it now matches main behavior and passes participant_ids through unchanged.
• Keeps cleanup hardening by deleting the agent env and archiving the thread before deleting the agent, avoiding FK cleanup failures after early test errors.

Tests

• go test ./...
• go vet ./...
• go test -tags "e2e svc_agents_orchestrator" -run '^TestNoDuplicateWorkloads$' ./tests attempted locally and reached the expected local environment boundary (dial agents:50051: context deadline exceeded) because the Kubernetes E2E stack is not running in this workspace.

[casey-brooks]

Update for #145

Implemented the thread contract and cleanup fixes.

Summary

• createThread now uses participants (ParticipantIdentifier) instead of deprecated participant_ids.
• createThread filters out the caller identity from initial participants, so the initiator is not sent in the participants list.
• TestNoDuplicateWorkloads cleanup now deletes the created agent env and archives the thread before deleting the agent, preventing FK cleanup failures after early errors.

Test & lint summary

• go test ./...: 0 failed; packages reported no test files.
• go vet ./...: passed with no errors.
• go test -tags "e2e svc_agents_orchestrator" -run '^TestNoDuplicateWorkloads$' ./tests: attempted locally; failed at environment boundary because the local Kubernetes E2E stack is not running (dial agents:50051: context deadline exceeded).

[noa-lucent]

Thanks for the update. I found one blocking issue in the shared thread helper: it silently proceeds when the initiator identity is missing from the outgoing context, which can reintroduce the invalid participant list. Please make that identity requirement explicit/fail-fast before merge.

[casey-brooks]

Update after review

Addressed Noa's strict initiator identity requirement.

Summary

• createThread now calls requireIdentityIDFromOutgoingContext, which fails loudly if outgoing identity metadata is missing or empty.
• The participant list is only built after a non-empty initiator identity is present, so the helper cannot silently send the initiator in the participants list when context metadata is missing.

Test & lint summary

• go test ./...: 0 failed; packages reported no test files.
• go vet ./...: passed with no errors.
• go test -tags "e2e svc_agents_orchestrator" -run '^TestNoDuplicateWorkloads$' ./tests: attempted locally; failed at environment boundary because the local Kubernetes E2E stack is not running (dial agents:50051: context deadline exceeded).

[rowan-stein]

Heads up: I merged threads PR that fixes the underlying contract (initiator auto-included + cascade deletes): agynio/threads#36. This may supersede the need to change go-core createThread behavior in e2e PR #146.

[noa-lucent]

Re-reviewed with the merged threads fix in mind. Since agynio/threads#36 now fixes the underlying CreateThread contract and cascade cleanup, the createThread compatibility/filtering change in this PR should be removed so the E2E suite exercises the fixed service behavior. Cleanup hardening in the dedup test can remain if still useful.

[casey-brooks]

Update after re-review

Adjusted PR #146 now that agynio/threads#36 is merged.

Summary

• Reverted the createThread helper changes so it matches main behavior again and passes participant_ids through unchanged.
• Removed the initiator filtering/strict metadata logic from the shared helper so E2E exercises the fixed threads service behavior directly.
• Kept the TestNoDuplicateWorkloads cleanup hardening: delete agent env, archive thread, then delete agent.

Test & lint summary

• go test ./...: 0 failed; packages reported no test files.
• go vet ./...: passed with no errors.
• go test -tags "e2e svc_agents_orchestrator" -run '^TestNoDuplicateWorkloads$' ./tests: attempted locally; failed at environment boundary because the local Kubernetes E2E stack is not running (dial agents:50051: context deadline exceeded).

[noa-lucent]

Re-review complete. The createThread workaround has been reverted, so E2E now exercises the fixed threads service behavior from agynio/threads#36. The remaining changes are limited to TestNoDuplicateWorkloads cleanup hardening, and CI is green on the current head SHA.