[kubernetes/container_logs]: Allow creating single input to monitor all log files

### Integration Name

Kubernetes [kubernetes]

### Dataset Name

kubernetes.container_logs

### Integration Version

1.84.1

### Agent Version

9.2.1

### OS Version and Architecture

Darwin 25.2.0

### User Goal

Add an opt-in config option to create one filestream input to monitor the entire cluster. The `paths` would be set to `/var/log/pods/*/*/*.log` by default.

This feature will be opt-in and will NOT be the default.

### Existing Features

The kubernetes.container_logs data stream creates one filestream input per container. For 100s of pod and potentially 1000s of containers, this becomes an issue because each input comes with a baseline memory usage.
This can lead to OOMs and restarts in production environment.

### What did you see?

I manually installed a policy with one filestream input monitoring `/var/log/containers/*.log` as a whole and tested it against the latest version kubernetes integration. That showed remarkable reduction in overall memory usage.

Attaching images:

1. Kubernetes integration 1.84.1.

2. Custom policy with one input and rest of the config same as previous one.

### Anything else?

If any customer wants to switch to this feature, they'd face data duplication as the input id will change. We'll be documenting this caveat.