Ordered processing pipeline

One manager processes accepted occurrences sequentially. A slow Reporter cannot move a later incident ahead of an earlier one, and two blocking presentations from the same manager cannot overlap.

capture
  → classify
  → collect context
  → plan actions
  → sanitize
  → report
  → present
  → recover
  → notify completion

ErrorPolicy produces one ErrorActionPlan with independent directives:

• reporting: skip or record;
• presentation: none, passive, or blocking;
• recovery: none, resetScope, resetApplication, or terminate.

The Sanitizer creates an ErrorReport before any Reporter runs. Presenter and Recovery remain inside the application and can inspect the raw ErrorIncident; a Reporter cannot.

Scheduling and synchronization

captureError schedules processing and returns whether the manager accepted the occurrence. handleError returns the operation Future. flush waits for queued work without closing the manager. close drains the queue and ends the lifecycle permanently.

Failures inside the pipeline

Each extension point is isolated. If one Reporter throws, later Reporters still receive the original sanitized report. The failure goes directly to ErrorPipelineFailureHandler; it is never resubmitted to the same manager as a new incident.