Add Helm chart for Kubernetes deployment

[megative]

What

Helm chart at helm/incidentrelay: three deployments (web, scheduler, telegram) from the single image switched by INCIDENTRELAY_SERVICE, Service + optional Ingress, /healthz//readyz probes with a startup probe covering entrypoint migrations, config rendered from values into a Secret (existingConfigSecret supported), SQLite on a PVC or external PostgreSQL. Deployments default to the Recreate update strategy to protect the shared SQLite volume during rollouts. README gets a "Kubernetes (Helm)" install section.

Known caveats are documented in values/NOTES/README: no public image is published yet (image.repository must point at your own build), SQLite is single-node only, and migrations run on web pod start so web.replicaCount should stay at 1 unless migrations are moved out of band.

Tests

No backend code is touched, so no tests are added: the change is deployment tooling only. Validated with helm lint / helm template across value combinations and on a live kind cluster: all three workloads become Ready, probes pass, an alert sent through a real Alertmanager webhook receiver was ingested via a route intake token and acknowledged/resolved, and the SQLite database survives pod recreation on the PVC.

[Aidaho12]

Hello, have you tried to run on kubernetes?

[megative]

Hi, just locally on a kind cluster. The web has started however I'd like to make some additional tests since I've found problem with alerting in current realization just after creating of PR. Let's keep it as draft for a while because I need couple of days to return to it.

I'll notify you additionaly.

[megative]

Retested the chart locally on kind after rebasing the branch on latest upstream/main.

• Installed the chart with a locally built incidentrelay:helm-test image.
• web, scheduler, and telegram deployments became Available.
• /healthz returned {"status":"ok"}.
• /readyz returned database ok with 23/23 migrations applied.
• Verified Alertmanager ingestion using the infra-alertmanager demo route token: the API created alert group id=1, status=firing, team=infra, route=infra-alertmanager.

During the manual UI check I found a separate frontend asset-path issue: the app was loading the alerts summary grid component from the wrong static path, so /api/alerts returned data but the Alerts page failed to render with renderAlertsSummaryGrid is not defined. I split that into a separate focused PR: #8

[Aidaho12]

It means PR is ready?

[megative]

Yep!