fix: clear error when triggering job on undeployed scheduled task

[fedemaleh]

Summary

Dedicated PR for the scheduled-task trigger fix, split out from #202 (which also carries the unrelated kill-job-execution work). No overlap with that work — this touches only trigger files.

Triggering the Trigger job action on a scope with no active deployment previously produced an opaque error: with no deployment there is no CronJob, so the script ran kubectl create job --from="cronjob/" (no name) and surfaced a cryptic kubectl failure.

Changes

• Guard on .scope.current_active_deployment — if empty/null, fail early (before any kubectl call) with a clear "deploy the scope first" message. Also guards the empty-CronJob case with its own actionable message.
• set -euo pipefail added (was missing), aligning with the repo's scope-script conventions.
• log function via workflow step — trigger-job.yaml now has a load logging step importing $OVERRIDES_PATH/logging (a scheduled-task copy of k8s/logging); the script assumes log exists rather than sourcing it, matching restart-pods.yaml.
• include: $SERVICE_PATH/values.yaml in trigger-job.yaml so K8S_NAMESPACE (and LOG_LEVEL) are provided, plus a script-level ${K8S_NAMESPACE:-nullplatform} guard as defense-in-depth (the missing include caused an unbound variable crash under set -u).

Test plan

New scheduled_task/scope/tests/trigger.bats — 10/10 passing:

• Happy path; no active deployment (key absent + null); no CronJob found
• Namespace resolution (provider namespace / default / K8S_NAMESPACE unset / nullplatform fallback)
• Workflow-wiring checks: values.yaml include present, load logging step present

bats scheduled_task/scope/tests/trigger.bats
1..10  ok 1..10