Summary

Add TryEachInContextAsync — a best-effort batch runner that continues processing remaining items when individual ones fail, collecting errors instead of aborting the whole batch.

Goal

Support partial success: keep 995 of 1000 emails when 5 fail, reporting failures at the end rather than halting.

Problem

ExecuteAll/Map propagates the first exception and abandons the rest. For bulk writes/notifications, a single bad address should not discard the remaining legitimate items.

Scope

• TryEachInContextAsync returning a result object.
• result record: SuccessCount, FailureCount, HasErrors, Failures ({Item, Exception}).
• Each item runs in a context snapshot; exceptions per item are caught and collected.
• Optional CancellationToken to abort between items.
• Unit tests: mixed success/failure scenarios.

Design Expectations

• Per-item isolation and events identical to base execution.
• Failures are structural (exception) not state-corrupting.
• Pairs naturally with MapParallel for bounded best-effort batches.

Acceptance Criteria

• Returns counts and failure list instead of throwing on first item.
• Does not corrupt context across items.
• Respects cancellation between items.

Non-Goals

• No retry (resilience lives in the Polly adapter).
• No ordering guarantee (see Ordered/Unordered).
• No automatic partial-failure rollback.