Runaway Query Kill Action Delayed Beyond Configured Threshold

## Bug Report

Please answer these questions before submitting your issue. Thanks!

### 1. Minimal reproduce step (Required)

### 2. What did you expect to see? (Required)

< We expected any query that exceeds the configured runaway query threshold of 300 seconds to be terminated immediately (or within a negligible delay) after reaching the threshold.

Since the resource group is configured with the action set to KILL and the threshold set to 300 seconds, we expected all matching queries to be consistently killed at around 300 seconds regardless of whether they are executed using TiKV or TiFlash/MPP.>

### 3. What did you see instead (Required)

### 4. What is your TiDB version? (Required)

Start by reproducing the runaway query rule with a 300-second KILL threshold on TiDB v8.5.3, comparing TiKV with TiFlash/MPP execution. Investigate the query-kill propagation timing and determine whether the roughly 419-second termination is expected; done means identifying the cause and making termination consistent near the configured threshold or documenting the expected delay.

Written by the indexing model from the issue text.