[fix] resolve {PREFIX} metric init order in module_manager

[diyliv]

Overview

Fix {PREFIX} placeholder not being replaced in two addon-operator metrics (mm_module_info and mm_module_maintenance), causing them to appear as _PREFIX_mm_* in Prometheus instead of deckhouse_mm_*.

What this PR does / why we need it

Package-level variables moduleInfoMetricName and moduleMaintenanceMetricName in module_manager.go copied metric names from the metrics package at Go init time, before InitMetrics() could replace the {PREFIX} placeholder. At call sites, these stale copies were used instead of the updated metrics.ModuleInfoMetricName / metrics.ModuleMaintenanceMetricName.

This caused the ModuleIsInMaintenanceMode Prometheus alert to never fire, because the alert expression referenced deckhouse_mm_module_maintenance while the actual metric was exported as _PREFIX_mm_module_maintenance.

The fix removes the two package-level variable copies and reads directly from metrics.ModuleInfoMetricName / metrics.ModuleMaintenanceMetricName at call time, after InitMetrics() has already replaced the prefix.

Special notes for your reviewer

Verified in a staging Deckhouse cluster — before the fix, only _PREFIX_mm_module_maintenance and _PREFIX_mm_module_info were present; all other metrics correctly used the deckhouse_ prefix.