The sidecar injected into pg_basebackup replica clusters ignored the
instanceSidecarConfiguration (resources, logLevel, additionalContainerArgs) of
the replica-source ObjectStore: collectSidecarResourcesForPod and
collectAdditionalInstanceArgs only handled the cluster and recovery object
stores. Log-shipping replicas masked this because their recovery and replica
source object stores coincide; pg_basebackup replicas only set
ReplicaSourceBarmanObjectName, so the configuration was silently dropped (the
sidecar ran with no additional args and the default log level).
Handle ReplicaSourceBarmanObjectName in both helpers, documenting the store
precedence (Barman > Recovery > ReplicaSource), and fix the misleading comment
on the recovery branch.
Signed-off-by: Armando Ruocco <armando.ruocco@enterprisedb.com>
