Report the admin API error when collecting group lags

[omlahore]

collectConsumerGroupLagsAdminAPI drops the error from ListAllConsumerGroupOffsetsAdminAPI:

groupOffsets, err := e.minionSvc.ListAllConsumerGroupOffsetsAdminAPI(ctx)
for groupName, offsetRes := range groupOffsets {

ListAllConsumerGroupOffsetsAdminAPI returns nil, fmt.Errorf("failed to list groupsRes: %w", err) when listConsumerGroupsCached fails (minion/consumer_group_offsets.go:22). Ranging over that nil map is a no-op, so the function falls through to return isOk with isOk still true.

The effect is that a failed admin call is reported as a successful scrape that happens to contain no consumer group lag metrics, rather than as a failure. For an exporter that is the wrong way round: alerts built on kafka_consumergroup_group_lag stop firing instead of the scrape going unhealthy.

err is also reassigned a few lines down inside the loop (err = kerr.ErrorForCode(...)), so the original value could not have been checked later even in principle. staticcheck flags it as SA4006.

The fix follows the shape already used in collect_broker_info.go:12: log with zap.Error and return false.

Verified: go build ./... passes, gofmt -l clean, and SA4006 is gone under both GOOS=linux and GOOS=darwin. prometheus/ has no test files, so there was nothing to extend.