Network and kubernetes instability in extended use sessions

### Description

If I leave my colima/kubernetes running for some time, it destabilizes unpredictably. The two signs of that happening are:

1. I start seeing network connectivity errors from various services
2. If I clean up all deployments and services, all pods get stuck in Terminating.

### Version

colima version 0.6.7
git commit: ba1be00e9aec47f2c1ffdacfb7e428e465f0b58a

runtime: docker
arch: x86_64
client: v24.0.7
server: v24.0.7
limactl version 0.19.1
qemu-img version 8.2.0
Copyright (c) 2003-2023 Fabrice Bellard and the QEMU Project developers

### Operating System

- [ ] macOS Intel <= 13 (Ventura)
- [ ] macOS Intel >= 14 (Sonoma)
- [ ] Apple Silicon <= 13 (Ventura)
- [X] Apple Silicon >= 14 (Sonoma)
- [ ] Linux

### Output of `colima status`

INFO[0000] colima is running using QEMU
INFO[0000] arch: x86_64
INFO[0000] runtime: docker
INFO[0000] mountType: sshfs
INFO[0001] socket: unix:///Users/aran/.colima/default/docker.sock
INFO[0001] kubernetes: enabled

### Reproduction Steps

I run a variety of services on Kubernetes and just let it sit. I use a tool called Skaffold to deploy to the local Kubernetes. It uses docker to load containers directly. When errors crop up everywhere, I kill Skaffold and it deletes the services and deployments. At that point I see all pods stay in Terminating status. If I restart colima, when it comes back up, those pods then finish Terminating normally.

### Expected behaviour

Pods retain connectivity and always terminate normally within colima kubernetes instance.

### Additional context

Kubernetes version v1.28.3+k3s2
M2, Sonoma 14.2.1, 24GB RAM
cpu: 3
disk: 60
memory: 16

El issue describe inestabilidad de red y de la terminación de pods en el runtime de Kubernetes de colima después de un uso prolongado. Empieza examinando el código fuente de colima para el manejo de la red y la gestión del ciclo de vida de los pods, especialmente en torno a la integración con k3s. Reproduce el problema dejando una instancia de colima en ejecución con deployments y, después, comprueba los logs y la conectividad de red. Busca tests existentes relacionados con la estabilidad de Kubernetes o con problemas de red.

Escrito por el modelo de indexación a partir del texto del issue.