CI: update tests assert codes, not sentences (#1089)

[Mikola Lysenko (mikolalysenko)]

LLM Description written by Claude Code:claude-opus-5-5

Refs #1089 (update slice of child 6; the issue stays open for the scan/ covgap files and the core crate's covgap_* files)

Summary

The three update/covgap_* files pinned English sentences, and their names kept "covgap" as a test category. This PR moves each test to a named home:

Old	New
covgap_commands_update.rs already-latest --json	update/update_json_contracts.rs: skipped/already_latest event, details.current/latest, non-empty reason, summary, warnings omitted, assert_envelope_invariants; the reason sentence assert is gone
covgap_commands_update.rs pin-to-current (human)	update_json_contracts.rs with --json: one skipped event, dryRun:false, zero network, notifier cache not written; plus one wording check in update/update_render.rs
covgap_commands_update.rs human dry-runs + PTY decline	update/update_render.rs, one check per message family
covgap_update_swap.rs (asserted Updated socket-patch)	update_json_contracts.rs with --json: downloaded + updated events (from/to), plus the same disk, inode and lock-file proofs
covgap_update_download.rs	self_update_failures_e2e.rs::asset_server_error_is_download_failed, next to its 404 sibling

DEAD_ENDPOINT moves into tests/common/update_fixture.rs. No product code changes.

CI cost: test-only. Lean PR run is unchanged: the same suites and job count, and the same test count (2 tests become 4 checks, all zero-network or local wiremock, under a second each).

Test evidence

• cargo test -p socket-patch-cli --all-features --test update --test self_update_failures_e2e --test self_update_e2e: 63 + 38 + 30 passed.
• cargo fmt --all -- --check, cargo clippy --locked --workspace --all-features -- -D warnings: clean.
• This PR moves tests, so there is no red→green regression test. The moved checks cover the same paths as before, and the pinned and unlocked-swap paths now also check their --json envelope.

/code-review high findings

Fixed: the bare "500" match (it could hit the mock's port), the missing reason check, the pinned test's missing dryRun/summary checks, DEAD_ENDPOINT placement, and the render-suite doc.
Not fixed:

• Pinned no-op reports already_latest with latest = <pin>. That is product behavior. This PR no longer asserts that code or field for the pinned path, so it doesn't lock them in. Whether the pinned envelope should use a different code is a product decision for a separate issue.
• The hand-written 500 wiremock mock should become a FakeReleaseBuilder option. It was moved verbatim from the covgap file. Adding a builder option would widen this PR for one caller.

🤖 Generated with Claude Code

https://claude.ai/code/session_017tesJSyAELgQJ9X4mF2mWn

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

BugBot review

---

Generated by Claude Code

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 9dd5b35. Configure here.}