Add further metrics to measure datastore accesses (per gRPC ops)

[alexsnaps]

Currently, we expose metrics using the prometheus crate.
We did add tokio-tracing to the mix recently and it would be great to consolidate the observability using the latter.

• Replace existing metrics with to leverage the tokio-tracing existing metrics
• Have aggregated metrics for all calls to the external datastore, per logical operation New time metric aggregation layer and refactor of prometheus metrics #267