feat(temporal): expose max_cached_workflows on AgentexWorker

Add a max_cached_workflows parameter (default 1000, unchanged) to AgentexWorker and pass it through to the temporalio Worker. The sticky workflow-cache size was previously not configurable, so agents with many concurrent long-lived workflows (e.g. ones that park waiting for events) could not raise it above the SDK default, causing heavy sticky-cache eviction and constant workflow replay under load.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: NiteshDhanpal

src/agentex/lib/core/temporal/workers/worker.py

@@ -151,6 +151,7 @@ def __init__(
         task_queue,
         max_workers: int = 10,
         max_concurrent_activities: int = 10,
+        max_cached_workflows: int = 1000,
         health_check_port: int | None = None,
         plugins: list = [],
         interceptors: list = [],
@@ -162,6 +163,7 @@ def __init__(
         self.activity_handles = []
         self.max_workers = max_workers
         self.max_concurrent_activities = max_concurrent_activities
+        self.max_cached_workflows = max_cached_workflows
         self.health_check_server_running = False
         self.healthy = False
         self.health_check_port = (
@@ -227,6 +229,7 @@ async def run(
             activities=activities,
             workflow_runner=UnsandboxedWorkflowRunner(),
             max_concurrent_activities=self.max_concurrent_activities,
+            max_cached_workflows=self.max_cached_workflows,
             build_id=str(uuid.uuid4()),
             debug_mode=debug_enabled,  # Disable deadlock detection in debug mode
             interceptors=self.interceptors,  # Pass interceptors to Worker