refactor(indexer): extract ScheduledTask abstraction from Dispatcher

What does this MR do and why?

The Dispatcher trait coupled two concerns: scheduling (cadence locks, intervals) and semantics (publishing work to NATS). Adding a scheduled task that doesn't fan out work — like a daily OPTIMIZE TABLE FINAL CLEANUP — meant either abusing the dispatcher abstraction or duplicating the scheduling infrastructure.

This MR extracts a ScheduledTask trait that only cares about when to run, not what the task does. The existing dispatchers implement it unchanged; they just happen to publish to NATS in their run() method. Future maintenance tasks can implement the same trait without pretending to be dispatchers.

Changes

Trait and type renames:

Before	After
Dispatcher	ScheduledTask
dispatch()	run()
dispatcher_config()	schedule()
DispatchError	TaskError
DispatcherConfiguration	ScheduleConfiguration
DispatchConfig	ScheduleConfig
DispatchMetrics	ScheduledTaskMetrics

Metrics:

• Moved ScheduledTaskMetrics from modules/sdlc/dispatch/ to dispatcher/ — it's shared infrastructure, not SDLC-specific
• Meter: indexer_dispatch → scheduler
• Metric names: indexer.dispatch.* → scheduler.task.*
• Label: dispatcher → task

Task names now include a category prefix (dispatch.sdlc.global, dispatch.code.project) so non-dispatch tasks added later are naturally distinguished.

Config keys: dispatch.dispatchers → schedule.tasks (helm values and templates updated).

Testing

Unit and integration tests are ok

Performance Analysis

• This merge request does not introduce any performance regression. If a performance regression is expected, explain why.