fix(codeapi-chart): support sandbox runner scale to zero

[rschlaefli]

What This Fixes

This PR adds a chart value for CodeAPI deployments where codeapi-sandbox-runner is scaled from zero by an external scaler such as KEDA.

By default the chart still renders the existing service-worker init container that waits for codeapi-sandbox-runner. For scale-to-zero deployments, set:

workerSandbox:
  serviceWorker:
    waitForSandboxRunner: false

That lets the service-worker start while the sandbox runner is at zero, so queued/active BullMQ state can wake the runner through the platform scaler.

Branch Coverage

• Base: main
• Head: dc1e56a
• Reviewed: 1 commit, 3 files changed.
• Covered: chart value default, conditional init container rendering, README deployment note.

Verification

Current head:

• GitHub Actions API Unit Tests on dc1e56a -> passed.
• GitHub Actions Service Unit Tests on dc1e56a -> passed.
• helm template codeapi helm/codeapi -n codeapi --set executionManifest.privateKey=dummy --set executionManifest.publicKey=dummy | rg "wait-for-sandbox-runner|wait-for-redis" -> default renders both init containers.
• helm template codeapi helm/codeapi -n codeapi --set executionManifest.privateKey=dummy --set executionManifest.publicKey=dummy --set workerSandbox.serviceWorker.waitForSandboxRunner=false | rg "wait-for-sandbox-runner|wait-for-redis" -> renders wait-for-redis only.
• git diff --check -> pass.

Security / Privacy

• Review: chart-only deployment behavior change.
• Result: no new images, secrets, RBAC, permissions, network paths, or runtime code.
• Sensitive data: no secret values included.

Blocking Before Merge

• Reviewer accepts the scale-to-zero chart value.

Follow-Up After Merge

• Update df-cloud CodeAPI target revision to the merged commit.
• Merge the helm-charts companion change that disables this wait and adds BullMQ active KEDA triggers.