closedts: understand spurious "closed timestamps side-transport connection dropped" errors

Across our fleet, we see[^1] a significant volume of these logs:

> closed timestamps side-transport connection dropped from node: 9 (grpc: ‹×› [code 1/Canceled])

These are likely spurious, and not caused by network issues.

We saw these errors in an escalation as well, during periods of time in which no nodes were restarting (which might otherwise be a likely source of those errors).

[^1]: https://us5.datadoghq.com/logs?query=service%3Acockroachdb%20%40channel%3A%28DEV%29%20%22closed%20timestamps%20side-transport%20connection%20dropped%22&agg_m=count&agg_m_source=base&agg_t=count&cols=host%2Cservice&fromUser=true&messageDisplay=inline&refresh_mode=sliding&storage=flex_tier&stream_sort=desc&viz=stream&from_ts=1759133960972&to_ts=1759134260972&live=true

Jira issue: CRDB-54887