chore(api): log reason for failing to connect, resume sandboxes on state change

[matthewlouisbrockman]

currently, we drop the err from err = a.orchestrator.WaitForStateChange(ctx, teamID, sandboxID). This makes it hard for us to know why requests during pausing errored.

This change logs the dropped err so we can get better confidence in why we're gettign 500 errors during state transitions.

also updates the error logging to use the telemetry.reportCriticalError to maintain Error text while helping with the spans

---

Note

Low Risk
Low risk: changes are limited to error-path telemetry/logging while keeping existing control flow and responses largely the same, with added visibility into WaitForStateChange failures and snapshot/token errors.

Overview
Improves observability for sandbox connect and resume during state transitions by reporting previously dropped WaitForStateChange errors and standardizing several failure logs to telemetry.ReportCriticalError/ReportErrorByCode with sandbox/team (and where relevant template/build) metadata, making 500s and other failures easier to diagnose without changing the happy-path behavior.

^{Written by Cursor Bugbot for commit b289483. This will update automatically on new commits. Configure here.}

[dobrac]

you can use the telemetry.ReportError and telemetry.ReportCriticalError instead to also populate the traces correctly. It does logging automatically

[matthewlouisbrockman]

you can use the telemetry.ReportError and telemetry.ReportCriticalError instead to also populate the traces correctly. It does logging automatically

shoud we add him to logger.L().Error(ctx, "Error getting last snapshot", logger.WithSandboxID(sandboxID), zap.Error(err)) too to debug those guys? (i think it's a 30s timeout somewhere but can't find it yet)

[matthewlouisbrockman]

eh, can worry about getting the reason for the failed db guys later, pretty sure those are all just timing out on the 30 second callback from the redis route

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}