Parallel Agent Execution overview

Parallel Agent Execution as a cross-cutting topic - dispatching multiple agents concurrently to work on independent tasks. Covers ctx.parallel.all() and ctx.parallel.map() for fan-out dispatch, worktree isolation to prevent merge conflicts during parallel file edits, result aggregation and quality scoring across parallel branches, synchronization barriers before dependent phases, and failure handling when one parallel branch fails while others succeed. Enables significant throughput gains for decomposable tasks like multi-file refactoring or independent feature implementation.