chore(release): cut orb-v0.5.0

[github-actions]

Proposes cutting orb-v0.5.0 (a minor release) from the current
stable release orb-v0.4.0.

Merging this PR tags orb-v0.5.0 and dispatches the image build/publish --
which still requires a release environment approval before anything reaches GHCR (see
.github/workflows/release-selfhost.yml). The daily beta channel is unaffected either way.

Image-relevant commits since the last stable release

• fix(selfhost): widen 2 more GitHub-id columns found via a live-schema sweep (fix(selfhost): widen 2 more GitHub-id columns found via a live-schema sweep #5062) (2205cde)
• fix(review): log a bounded response snippet on unparseable AI output (ai_review_provider_unparseable_exhausted: no visibility into WHY the model's output didn't parse #5071) (fix(review): log a bounded response snippet on unparseable AI output #5073) (500ddad)
• fix(review): a single oversized/malformed chunk no longer fails the whole embedding batch (embedTexts: one oversized/malformed chunk fails the WHOLE embedding batch, not just that chunk #5072) (fix(review): stop one oversized RAG chunk from failing its whole embedding batch #5074) (ad44723)
• fix(linear): stop fuzzy matching workspace milestones (fix(linear): stop fuzzy matching workspace milestones #4977) (66ed61d)
• fix(commands): give ask/chat their own fresh reply comment (fix(commands): give ask/chat their own fresh reply comment #5077) (99c2429)
• fix(selfhost): stop claude-code from treating its own review call as an interactive plan or a prompt injection (claude-code reviewer runs in interactive Plan Mode and treats its own review instructions as a prompt injection #5079) (fix(selfhost): stop claude-code from treating its own review call as an interactive plan or a prompt injection #5080) (6784974)
• feat(review): add a [BETA] chat usage collapsible to the PR comment (feat(review): add a [BETA] chat usage collapsible to the PR comment #5081) (74afe7d)
• feat(review): fleet-wide gaming-pattern detector for the self-host orb fleet (maintainer: fleet-wide anti-farming detector — flag a miner/fleet gaming the calibration score #2350) (feat(review): fleet-wide gaming-pattern detector for the self-host orb fleet (#2350) #5083) (1822fca)
• fix(stats): scope reuse trend to public repos (fix(stats): scope reuse trend to public repos #4991) (f22516c)
• fix(selfhost): retry the /ready probe in selfhost-post-update-check.sh instead of a single attempt (selfhost-post-update-check.sh false-fails on a normal deploy: single /ready probe, no retry #5085) (fix(selfhost): retry the /ready probe in selfhost-post-update-check.sh instead of a single attempt #5086) (0800f23)
• feat(commands): let a PR's own author use chat when rate limiting is active (feat(commands): let a PR's own author use chat when rate limiting is active #5087) (9b307d8)
• fix(review): gate the auto-tune breaker on reversal-weighted precision (maintainer: value-weighted gate-prediction calibration — optimize for merged-net-positive, not raw volume #2348) (fix(review): gate the auto-tune breaker on reversal-weighted precision (#2348) #5088) (978417e)
• fix(selfhost): retry the container health-status check, not just /ready (selfhost-post-update-check.sh: the container health-status check still isn't retried (follow-up to #5085) #5089) (fix(selfhost): retry the container health-status check, not just /ready #5090) (e2bc19b)
• fix(commands): require an open, non-draft PR for chat's pr_author grant (fix(commands): require an open, non-draft PR for chat's pr_author grant #5094) (eb3d641)
• fix(observability): fix Sentry fingerprinting for 3 fragmented issue groups (Fix Sentry fingerprinting for 3 fragmented issue groups (analyzer_timeout, AI-review-notes, publish-failed) #5010) (fix(observability): fix Sentry fingerprinting for 3 fragmented issue groups #5099) (86370f2)
• feat(ops): add fatigue-controlled PagerDuty paging for ops anomalies (Wire on-call/paging for the central hosted service #4937) (feat(ops): add fatigue-controlled PagerDuty paging for ops anomalies #5116) (c5b3808)

[superagent-security]

Superagent didn't find any vulnerabilities or security issues in this PR.