Add configurable retry policy for GCS integration connection failures

[cpascale43]

Currently, when the GCS integration encounters connection issues, it has a fixed retry policy of 2 retries (3 total attempts) for file downloads. This is not configurable by users, which can lead to integration failures when encountering temporary network instability.

Example error:

elastic_agent.filebeat [elastic_agent.filebeat][error] Input 'gcs' failed with: input.go:130: input gcs-google_cloud_storage failed: read tcp: connection reset by peer

Investigation is needed to:

• Determine if adding configurable retry settings is possible
• Assess which retry parameters could be exposed/which aspects of retry behavior are controlled by Elastic vs. GCS

See: https://github.com/elastic/enhancements/issues/22943