TERRA-43's chokepoint coverage claim outlives the chokepoint: TERRA-52 makes logger.ts a non-sole writer

TERRA-52 adds a second appEvents writer (the CTE INSERT that is atomic with the isConfirmedByClient UPDATE). logger.ts stops being the sole writer the moment that ships. TERRA-43 documented probe marking at logger.ts:33-39 as a CHOKEPOINT guarantee - every terra appEvents row passes through this one function, so a probe against any arm gets marked without that arm knowing the concept exists. That justification dies with a second writer. WHAT IS AND IS NOT BROKEN. TERRA-52 was ruled PROBE-FREE (audit BLOCK on audit-terra-ca-msvshkwvdulj, adopted by coder-terra-cc): the action does not read the probe header, does not transform event values, and copies no marker logic - because a request that flips isConfirmedByClient persists a REAL catalog decision, so there is no such thing as a probe on that path. That makes the new writer structurally incapable of emitting an unmarked probe row. So TERRA-43's claim is still TRUE - and it is true for a different reason than the one written down. THAT IS THE DEFECT THIS ROW CARRIES. The guarantee has silently moved from a property of the FUNNEL to a property of EACH WRITER, and nothing in the tree says so. A third writer added later inherits nothing: it will be reviewed against a comment claiming a chokepoint that no longer exists, and it may well have a legitimate probe path. The failure mode is exactly the one logger.ts:33-39 predicts in its own words - an arm nobody remembered to plumb writes an UNMARKED probe row. DELIVERABLE: correct TERRA-43's coverage claim where it is written (logger.ts comment and any doc restating it) so it states the real bound - marking is guaranteed for rows written THROUGH logger.ts, and every other writer must independently establish either that it marks or that it cannot produce a probe. Name the probe-free ruling as TERRA-52's discharge of that obligation. Class-A: touches the audit/telemetry trail's provenance claim. DO NOT MERGE WITH TERRA-53. That row is the marker being structurally UNAVAILABLE on non-error rows even when the logic runs correctly (markSynthetic writes errorMessage; detail.synthetic is explicitly not the marker). This row is about the coverage CLAIM resting on a funnel that no longer exists. Different defect, different fix.