refactor: reduce lint complexity

[mariusvniekerk]

Summary

• enable gocognit and gocyclo in golangci-lint with threshold 20
• split high-complexity Go functions into smaller helpers across CLI, GitHub sync, server, workspace, terminal, testutil, and tooling code
• update workspace delete/runtime cleanup flow to preserve dirty-delete sessions while keeping forced deletes safe

Test Plan

• make lint
• make test

[roborev-ci]

roborev: Combined Review (aa23ec9)

Medium risk: GraphQL sync failure handling can be masked by REST fallback.

Medium

• internal/github/sync.go:1512 - When GraphQL fetching succeeds but doSyncRepoGraphQL fails, the refactor returns false, falls back to REST, and can clear the failure if REST succeeds. Previously this path marked the MR sync scope as failed and skipped REST fallback, preserving force-refresh/backoff behavior. The same regression exists for issues in indexSyncRepoIssuesGraphQL.

  Fix: Distinguish “GraphQL unavailable/fetch failed” from “GraphQL sync failed”; return an explicit failure flag or error so failedScope is preserved.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)

[roborev-ci]

roborev: Combined Review (7b2426d)

High-level verdict: Changes need follow-up before merge due to workspace deletion race risk and missing required e2e coverage.

High

• internal/server/huma_routes.go:3317
BeginStopping was moved inside the Delete callback, but workspace.Manager.Delete only invokes that callback after preflight/dirty checks. The stopping guard no longer covers the full delete operation, so a concurrent launch can pass the runtime guard during deletion and race with StopWorkspace or worktree cleanup.
  Fix: Start BeginStopping before calling workspaces.Delete and defer EndStopping after it returns, or move the manager callback earlier so the guard covers the full delete flow.

• internal/github/sync_test.go:4226
  The GraphQL sync retry-state fix only adds unit coverage. This changes core sync/data-flow behavior and failure recovery, and the repository guidelines require e2e coverage for bug fixes in these paths.
  Fix: Add an e2e/API+SQLite full-stack test that exercises a GraphQL PR or issue sync failure and verifies the repo failure state is preserved for the next sync.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)

[roborev-ci]

roborev: Combined Review (61e91fc)

Code review verdict: no Medium, High, or Critical findings were reported.

All reviewed agents found the code clean for reportable issues.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)

[roborev-ci]

roborev: Combined Review (06a3440)

No Medium, High, or Critical findings were reported across the reviews.

All agents reported the code as clean or provided no findings.

---

Synthesized from 3 reviews (agents: codex, gemini | types: default, security)