Celery Logs are not respecting log format on stack traces

## Describe the bug

When celery logs an error with a stack trace, the error is not correctly consumed by the default python multiline parser in fluentbit. All other applications work as expected with the python parser.

## [Bug Severity]

Sev-2

## To Reproduce

- Disable custom celery mutliline parser in fluentbit config
- Redeploy/restart fluentbit to a scratch environment
- Break the signing key in celery to force stack trace errors
- View cloudwatch to see the errors split out over multiple log entries

## Expected behavior

- Celery logs work with the default fluentbit python multiline parser

## Impact

Impact on Notify users: None

Impact on Recipients: None

Impact on Notify team:
Increased maintenance and resource usage on fluentbit in Kubernetes.

Reproduce the issue by disabling the custom Celery multiline parser in the Fluent Bit configuration, restarting Fluent Bit in a scratch environment, and forcing a signing-key failure. Compare the resulting CloudWatch entries with the expected default Python multiline-parser output; done means the stack trace remains one correctly formatted log entry.

Written by the indexing model from the issue text.