Insanly huge agent cpu spike

**Describe the bug**
Hi,
I am encountering huge CPU usage of the Micro ros agent running from docker as follows:
`docker run -it --rm -v /dev:/dev --privileged --net=host microros/micro-ros-agent:foxy udp4 -p 8888`

**To Reproduce**
Steps to reproduce the behaviour:
1. Clone repo 'micro_ros_arduino' Foxy Branch
2. Flash 8 ESPs with it (Client Key unique for each Esp)
3. Run Publisher example for 3 hours
4. CPU spike happens after around 3h
5. See CPU spike on htop

**Expected behaviour**
A normal usage of ~100% CPU usage max at all times.

**System information (please complete the following information):**
- OS: Ubuntu 20.04
- ROS 2 Foxy
- Hardware: 8x ESP32 micro-controllers running agent and client version 2.3.0

what can be the possible reason behind such a huge CPU usage?
Observation from my tests: it happens exponentially were it starts to increase to 200 then 500 then 1000%
And after the agent reaches past 200%, the Esp32 loses connection with the agent and fails to publish any messages.
Any help is much appreciated!

Start by running the provided Docker command on the reported Ubuntu 20.04 and ROS 2 Foxy setup with eight ESP32 clients, then monitor htop during the Publisher example. Reproduce the delayed CPU growth and connection loss, identify the agent behavior responsible, and verify that CPU usage remains stable while all clients continue publishing.

Written by the indexing model from the issue text.