The coverage collector in openstack-network-exporter currently only emits metrics that are reported by OVS's `coverage/show` command. However, OVS only reports coverage counters that have been incremented (non-zero values), which means many defined metrics are never exposed to Prometheus until an error/event first occurs.

• Example: `ovs_coverage_drop_action_stack_too_deep_total` doesn't appear in Prometheus until the first occurrence

Expected Behavior:

• All  coverage metrics should be emitted, with zero values for counters not reported by OVS
• This matches the behavior of the pmd-perf collector where metrics like `ovs_pmd_lost_upcalls` correctly show zero values

Root Cause:
The `Collect()` function in `collectors/coverage/collector.go` only iterates over lines from OVS output, rather than iterating over all defined metrics in the `metrics` map.