0751 — Results follow maintained scenario families

Context

The user removed the historical experiment archive and clarified that results for current scenario families should remain. This supersedes 0750’s evidence-retention policy; its active workload selection and retired executable decisions continue.

Decision

• Keep quiet, max-throughput, older multitopic results belonging to the current multistream family, and published throughput evidence.
• Rename the dev result directory to quiet. Preserve original record contents, including declarations, durations, scenario names, and measurement identities. Reports select the containing directory; new quiet runs append there.
• Delete throughput exploration results: short automatic-batching tuning, distinct from the maintained explicit-batch max-throughput workload.
• Delete check-* and _controls: checker-validation and investigation artifacts, not maintained workload histories. Delete retired-source, fingerprint, and statistics scratch files as well.
• Do not recreate the historical experiment results the user deleted.

Consequences

Kept results follow scenario families even when historical names differ. just reliability-report quiet includes former dev records without rewriting those records as hour-long measurements. Retired experiment results are no longer available from the working tree; current documentation reflects that.