.NET: fix(dotnet): refresh reused workflow executor state - #9394

Motivation & Context

Read executor state from the current workflow context when a hosted workflow reuses an executor instance. Each run should load its persisted workflow state rather than retain an earlier run's cached value.

Description & Review Guide

• What are the major changes? Remove executor-local state caching and delegate reads and updates to the existing workflow-context state helpers. Preserve explicit full-executor reset opt-in and existing built-in reset declarations. Initialize state only when needed and queue the same value returned to the caller so nested and repeated reads stay consistent. Document the state and reset contracts, retain skipCache for API compatibility, and add hosted-session, interrupted-output, value-state, nested-read, repeated-initialization, and mixed-workflow reuse regressions. Make the test runner context's read-or-initialize implementation use its existing state manager rather than always returning new initial state.
• What is the impact of these changes? Reused executors read current scoped state, including pending updates and restored checkpoints, without opting subclasses' other instance fields into reset support. Nested initialization reads observe the same state object as the surrounding invocation. ChatProtocolExecutor uses the same context-scoped state for interrupted turns. Reset does not clear persisted workflow state. The executor-local cache optimization is removed.
• What do you want reviewers to focus on? Scoped state and checkpoint restoration, pending updates within a run, unchanged full-reset opt-in, and existing agent-host and handoff behavior.

Commands executed from dotnet:

• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --filter-class '*WorkflowExecutorStateTests' --no-progress: six original regression cases failed for the intended behavior before the fix; all fourteen final cases passed after the fix.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net472 --no-restore --filter-class '*WorkflowExecutorStateTests' --no-progress: passed, all fourteen cases. CI previously found two unsupported ValueTask factory calls in test helpers; these now wrap compatible canceled and faulted tasks without changing the assertions.
• dotnet build tests\Microsoft.Agents.AI.Workflows.UnitTests\Microsoft.Agents.AI.Workflows.UnitTests.csproj --no-restore --tl:off -v:quiet: passed across all target frameworks, zero warnings and errors.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --filter-method '*OutputExecutor_ResetClearsMessagesAfterInterruptedTurnAsync' --no-progress: two additional regression cases failed before the output executor reset repair; both passed afterward.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --filter-method '*MixedWorkflow_RequiresExplicitResetSupportForReuseAsync' --no-progress: failed before preserving explicit reset opt-in; passed after the revision.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --filter-method '*StatefulExecutor_PreservesNestedReadUpdatesDuringInitializationAsync' --no-progress: failed before the initialization repair because the nested update was lost; passed afterward.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --filter-method '*StatefulExecutor_InitializesOnlyWhenStateIsMissingAsync' --no-progress: both pending and committed state cases failed before deferring factory execution until initialization; both passed afterward.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.UnitTests -f net10.0 --no-restore --no-progress: passed, 858 tests. Seven existing tests initially failed after cache removal because the test runner context reinitialized state on every read; all passed after aligning that fixture with the state manager, with assertions unchanged.
• dotnet test --project tests\Microsoft.Agents.AI.Workflows.Declarative.UnitTests -f net10.0 --no-restore --no-progress: passed, 1,028 tests.
• dotnet build src\Microsoft.Agents.AI.Workflows\Microsoft.Agents.AI.Workflows.csproj --no-restore --tl:off -v:quiet: passed across all target frameworks, zero warnings and errors.
• dotnet format src\Microsoft.Agents.AI.Workflows\Microsoft.Agents.AI.Workflows.csproj --no-restore: passed.
• dotnet format tests\Microsoft.Agents.AI.Workflows.UnitTests\Microsoft.Agents.AI.Workflows.UnitTests.csproj --no-restore: passed.
• dotnet build agent-framework-dotnet.slnx --tl:off -v:quiet: blocked by unrelated package restore errors (NU1102); the configured feed does not contain A2A or A2A.AspNetCore version 1.0.0-preview3. The full-solution unit suite was not run.

Related Issue

N/A - maintainer-requested bugfix tracked internally; no public issue was opened. No matching open PR was found.

AI Assistance

• No material AI assistance was used.
• This is an AI-assisted contribution. I reviewed, understood, and verified all submitted content and accept responsibility for it.

AI assistance details: GitHub Copilot assisted with implementation, regression tests, XML documentation, and this PR description.

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR links to an agreed issue with no competing open PR, or the Related Issue section documents a trivial-change or repository-automation exception.
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and title prefix in sync automatically.