[Palo Alto Next-Gen Firewall]: Traffic logs with Link Switches field parsed incorrectly.

### Integration Name

Palo Alto Next-Gen Firewall [panw]

### Dataset Name

panw.panos

### Integration Version

5.4.0

### Agent Version

9.2.2

### Agent Output Type

elasticsearch

### Elasticsearch Version

9.2.1

### OS Version and Architecture

Elastic Cloud Hosted (AWS ARM)

### Software/API Version

PAN-OS 11.1.10

### Error Message

'100' is not an IP string literal.

### Event Original

<14>Dec 10 00:00:00 hostname 1,2025/12/10 00:00:00,000000000000,TRAFFIC,end,1,2025/12/10 00:00:00,10.0.0.0,10.0.0.0,0.0.0.0,0.0.0.0,Rule,,,x,x,x,x,x,x,x,2025/12/10 00:00:00,1,1,1,1,0,0,0x10001c,tcp,allow,1,1,1,1,2025/12/10 00:00:00,0,x,,0,0x8000000000000000,x,x,,1,1,tcp-fin,1,1,1,1,,hostname,from-policy,,,0,,0,,x,0,0,0,0,x,0,0,Policy ID,{link name, link tag, link type, physical interface, timestamp,100,100,,N/A},x,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,2025-12-10T00:00:00.000+11:00,,,x,x,x,1,x,,web-browsing,no,no,0,x,

### What did you do?

Integration configuration is default. Firewall configuration is close to default.

### What did you see?

See error message

### What did you expect to see?

Correct parsing.

### Anything else?

The issue is the CSV processor in the `logs-panw.panos-5.4.0-traffic` pipeline. It expects the log to be a comma-delimited list but the link_switches field (when present) is formatted as `{link name, link tag, link type, physical interface, timestamp, bytes read, bytes written, link health, link flap cause}` which is not recognised as a single field, but as 7 fields, meaning every field after that is matched to the wrong data. This eventually leads to the data from the `bytes written` sub-field being misinterpreted as the `xff_ip`, which causes an error that halts the pipeline

A similar issue occurs with the `characteristic_of_app` field, which is also a comma-delimited list.

Hopefully this can be easily fixed by replacing the csv processor with a grok processor.