fix: align metrics service names and reuse object metadata

[aslafy-z]

This change fixes inconsistent service naming and metadata handling across the metrics reconciler functions in fluentd, fluentbit and syslogng. It also refactors the code to reuse object metadata to ensure consistent behaviour in both present and absent states.

• For all three components (Fluentd, Fluentbit, SyslogNG), metrics-related services and buffer-metrics services now use a single objectMetadata variable initialized at the top of each reconciler function. This replaces repeated calls to *ObjectMeta(...) when building the Service or ServiceMonitor objects.
• This ensures that when disabling metrics or buffer-volume metrics, the correct Service/ServiceMonitor (with the "-metrics" or "-buffer-metrics" suffix) is targeted for deletion. Previously, the “absent” branch sometimes used a legacy suffix (e.g. "-monitor"), causing the actual metrics Service to be orphaned — this could break metrics cleanup and cause stale or duplicated resources.
• This resolves bugs where metrics services were not cleaned up or incorrectly named after disabling metrics or buffer-volume metrics, and helps avoid resource accumulation or inconsistency in deployments using any of fluentd, fluentbit or syslogng components.

[csatib02]

LGTM!