Enable retry behavior on swh rpc client connections

This should decrease the number of crashes when connection dropped because of a redeployment in swh rpc services.

It's already deployed for the scheduler rpc (directly in the code). This can be deployed through configuration. Adding enable_requests_retry: true to cls: remote configuration.

plan:

• swh/infra/ci-cd/swh-charts!512 (closed): Deploy in staging (& next version)
• Check behavior
• If ok, align in production