Task: Add scheduling support

Build the ores.scheduler library wrapping pg_cron with a Quartz.NET-style fluent API.

• Domain types: cron_expression (croncpp-backed, std::expected errors), job_status, job_definition, job_instance.
• Fluent builder: job_definition_builder with with_name / with_command / with_cron_schedule.
• cron_scheduler service: invokes cron.schedule / cron.unschedule via sqlgen.
• ores_scheduler_job_definitions_tbl hand-crafted (cron_job_id bigint, is_active integer don't fit the standard codegen templates); linked to cron.job via cron_job_id.
• Job instances read directly from cron.job_run_details via raw SQL join (no mirror table needed in Phase 1).
• description + is_active columns included for the future UI.
• cron_expression bridges pg_cron's 5-field Unix format and croncpp's 6-field format transparently.

Captured during execution; cleared into the parent story on close.

Quartz.NET pattern with all state in the DB: job_definition = persistent plan, job_instance = historical execution.

Scheduling library available end-to-end.