pub fn read_all_telemetry(
paths: &RunPaths,
) -> Result<Vec<(NodeId, TelemetryView)>, TelemetryError>Expand description
Read every projected node’s telemetry under one shared lock and one canonical-currency check. This is the bounded read-surface API for callers that need per-run rows or counts; it avoids recursively locking and rescanning the event log once per node.
A crash-tailed canonical projection cannot prove current attempts and
returns TelemetryError::RunStateNotCurrent. Malformed bounded sample
contents still classify as invalid; canonical projection, path-integrity,
and operational I/O errors propagate instead of masquerading as sample
corruption.