Store metrics queueTime in seconds

[ssncferreira]

What this PR does / why we need it:

metrics.go statistic queueTime was introduced in #4949. In this PR, the metric was stored in nanoseconds as int64:

loki/pkg/logqlmodel/stats/stats.pb.go

Line 102 in cecadf2

QueueTime int64 `protobuf:"varint,6,opt,name=queueTime,proto3" json:"queueTime"`

However, the statistics result struct:

loki/pkg/logqlmodel/stats/stats.pb.go

Lines 30 to 34 in cecadf2

	type Result struct {
	Summary Summary `protobuf:"bytes,1,opt,name=summary,proto3" json:"summary"`
	Querier Querier `protobuf:"bytes,2,opt,name=querier,proto3" json:"querier"`
	Ingester Ingester `protobuf:"bytes,3,opt,name=ingester,proto3" json:"ingester"`
	}

is used both for internal calculations and for API responses. Grafana uses these returned statistics from the API and expects time values to be returned in seconds as float. This PR converts the queueTime metric to be stored in seconds as float.
By storing time values in seconds as float64, the minimum value second we can store is 2.2E-308 (range for float64), which is 2.2E-299 nanoseconds. This means that we won’t lose precision when logging values less than a second or when sending the API response.

Special notes for your reviewer:

Checklist

• Documentation updated
• Tests updated
• Add an entry in the CHANGELOG.md about the changes.

[dannykopping]

LGTM 👍
Left a non-blocking question; if you'd prefer to not implement what I suggested, let me know and I'll merge