fix: activate flow schedules in the Scheduler Service (not just write the config)

[devin-ai-integration]

Summary

kbagent flow schedule created a keboola.scheduler config with state=enabled but never registered it with the Scheduler microservice, so the schedule was never activated and never fired — even though it looked "enabled" in schedule list/detail (those read the config field, not the Scheduler Service). The Keboola UI does two steps: write the config and activate it. We only did the first.

In Keboola, a keboola.scheduler config and an active schedule are distinct: the cron engine lives in a separate Scheduler service (scheduler.{stack}) that only runs schedules activated via POST /schedules {configurationId} (which also mints the schedule's token). This PR adds that activation step (and the symmetric deactivation on removal).

client.py — new Scheduler Service sub-client (sibling scheduler. host, same X-StorageApi-Token, same retry/backoff as queue/query/encryption)

activate_schedule(configuration_id, configuration_version_id=None) -> dict   # POST /schedules
list_activated_schedules(configuration_id=None) -> list[dict]                 # GET  /schedules (client-side filter)
deactivate_schedule(schedule_id) -> None                                     # DELETE /schedules/{id}

flow_service.set_flow_schedule — activate after writing the config

result = client.create_config(...) or client.update_config(...)
if effective_branch is None:                      # production only
    schedule = client.activate_schedule(result["id"], configuration_version_id=result["version"])
# result now carries: activated: bool, activated_schedule_id: str | None

Activation is production-only. On --branch, the config is written but not activated (activated=false); dev-branch schedules activate on deploy to production, mirroring how the Scheduler Service operates on the default branch.

flow_service.remove_flow_schedule — deactivate before deleting

For each matching keboola.scheduler config, delete its Scheduler Service activation first, so removal no longer orphans an activation + token that would keep firing against a deleted config. Result gains deactivated_schedule_ids.

CLI output

flow schedule now prints whether the schedule was activated (or that a dev-branch schedule activates on deploy).

Tests

• test_client.py::TestSchedulerService — activate (with/without version), list+filter, deactivate hit the correct scheduler. host/method/body.
• test_flow_service.py — activation happens on production, is skipped on a dev branch, and removal deactivates only the matching registration.
• Full suite: 4226 passed, 8 skipped (ruff, ruff format, ty, changelog-check all clean). Version bumped 0.66.0 → 0.66.1 with changelog + version-sync; gotcha + commands-reference updated.

Note

I couldn't exercise this against a live stack (no scheduler-service round-trip in unit tests). The endpoint contract (POST/GET/DELETE /schedules, configurationId = the keboola.scheduler config id) should be confirmed against the Scheduler API before/at merge, or via a quick manual flow schedule on a real project.

Link to Devin session: https://app.devin.ai/sessions/b4ab934128864f6d9e4b841e828bcc9e

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring