#1919 Telemetry flush failed (network error) warning appears repeatedly across multiple users/environments

Description

After nearly every CLI invocation (model method runs, workflow runs, report gets, etc.), the CLI prints:

2026-08-31T20:23:39.977Z [WRN] swamp·cli: Telemetry flush failed (network error) — entries are queued locally and will retry on the next invocation

Commands themselves complete successfully — this appears purely on the telemetry side-channel — but the warning is noisy and shows up on essentially every invocation, not as a one-off. Multiple users on my team have independently reported seeing this same warning, so it doesn't look isolated to one machine/network.

Steps to reproduce

1. Run any swamp command that talks to the CLI telemetry endpoint (e.g. swamp model method run <model> <method>, swamp workflow run <name>, swamp report get ...).
2. Observe the [WRN] swamp·cli: Telemetry flush failed (network error) line on stderr after the command's actual output.
3. Repeat on a subsequent invocation — the warning recurs; queued entries don't appear to successfully flush.

Expected behavior

Either telemetry flushes successfully in the background, or, if the endpoint is genuinely unreachable (proxy/firewall/etc.), the CLI should surface an actionable diagnostic (why it failed, how to disable telemetry persistently) rather than an unconditional recurring warning with no resolution path. Today --no-telemetry only silences a single invocation, and there's no persistent opt-out surfaced by swamp telemetry (only swamp telemetry stats exists).

Environment

• swamp version: 20260831.202137.0-sha.7401da9d
• OS: macOS (Darwin 25.5.0, arm64)
• Authenticated: yes (swamp-club.com)
• Reported independently by multiple users on the same team, not one machine