Fix: Prevent all servers from being deleted during second latency test

## Problem When multiple profiles have 'Auto-remove failed latency test servers' enabled, a critical bug causes all servers to be removed during the second latency test. ## Root Cause 1. After removeLatencyError() deletes failed servers, test queues still contain tags of deleted servers 2. When testing deleted servers, getByTag() returns null, causing immediate return and rapid scheduling loop 3. All servers are incorrectly removed due to incomplete testing ## Solution 1. Clean up test queues (testLatency/testLatencyIndepends) after removing servers in schedulerTestLatency() 2. Add 100ms delay before scheduling when encountering invalid servers to prevent rapid loop ## Impact - Fixes critical bug where all servers are lost - Maintains test integrity for valid servers - Fully backward compatible