kvclient: default distsender backoff is large for modern clusters

**Describe the problem**

The default retry options are:

```golang
// DefaultRetryOptions should be used for retrying most
// network-dependent operations.
func DefaultRetryOptions() retry.Options {
// TODO(bdarnell): This should vary with network latency.
// Derive the retry options from a configured or measured
// estimate of latency.
return retry.Options{
InitialBackoff: 50 * time.Millisecond,
MaxBackoff: 1 * time.Second,
Multiplier: 2,
}
}
```

The initial backoff may be an > 10x the RTT in many clusters. This makes issues like #173342 very impactful. We should revisit this TODO.

Jira issue: CRDB-66685

Start at the DefaultRetryOptions function shown in the issue and trace how distsender uses the returned retry.Options. Investigate the related behavior in issue #173342 and determine how configured or measured network latency should affect the initial backoff. Done means the retry behavior is appropriate for modern cluster RTTs, with tests covering the revised defaults or latency-dependent behavior.

Written by the indexing model from the issue text.