feat(rpc): measure request/call time in seconds

[shekhirin]

Updated dashboard

Changes

1. Show Active connections instead of Active requests, because requests may end up without responses hence without symmetrical finished metric. In that case we'll have this metric > 0, even though currently no requests are in processing.
2. Add Maximum call latency per method chart to help figure out what methods take too long.
3. Fix all heatmaps.

RPC duration reported in seconds instead of milliseconds

ubuntu@bench-box:~/reth-mainnet$ cast bn
18528468

ubuntu@bench-box:~/reth-mainnet$ curl -s localhost:9001/metrics | grep 'blockNumber'
reth_rpc_server_calls_failed{method="eth_blockNumber"} 0
reth_rpc_server_calls_successful{method="eth_blockNumber"} 1
reth_rpc_server_calls_started{method="eth_blockNumber"} 1
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0"} 0.000035252
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0.5"} 0.0000352551112514506
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0.9"} 0.0000352551112514506
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0.95"} 0.0000352551112514506
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0.99"} 0.0000352551112514506
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="0.999"} 0.0000352551112514506
reth_rpc_server_calls_time_seconds{method="eth_blockNumber",quantile="1"} 0.000035252
reth_rpc_server_calls_time_seconds_sum{method="eth_blockNumber"} 0.000035252
reth_rpc_server_calls_time_seconds_count{method="eth_blockNumber"} 1

[mattsse]

will a <1s call show up as 0?

[shekhirin]

@mattsse nope it's fine, see logs in the PR description

[mattsse]

lgtm