Add user latency tracking

[stevenlanders]

Notes:

• New flag --track-user-latency will track user latency metrics.
• It samples the latest block every --stats-interval and calculates the latency in that block.
• The latency is the block timestamp minus the value in the tx (value is set to time in epoch).
• The resolution here is seconds because block timestamps are in seconds. In general if our timing is subsecond we're okay.
• A lower --buffer-size can keep this lower.
• If this value is high, it could be that things are sitting in the mempool longer than they should.

2025/08/04 09:37:18 ============================================================
2025/08/04 09:37:28 throughput tps=152.30, txs=1523,  latency(avg=7ms p50=4ms p99=32ms max=87ms)
2025/08/04 09:37:29 user latency height=6908 txs=21 min=0s p50=0s max=0s
2025/08/04 09:37:38 throughput tps=104.40, txs=2567,  latency(avg=10ms p50=6ms p99=33ms max=111ms)
2025/08/04 09:37:38 user latency height=6940 txs=57 min=0s p50=0s max=0s
2025/08/04 09:37:48 throughput tps=166.30, txs=4230,  latency(avg=6ms p50=5ms p99=30ms max=111ms)
2025/08/04 09:37:48 user latency height=6977 txs=15 min=1s p50=1s max=1s

[pompon0]

?

[pompon0]

utils.Recv(ctx, ticker.C)

[pompon0]

there is no need to stop the ticker - it is GCed correctly since go 1.23 I think

[pompon0]

seems unused