Memory ingestion pipeline broken for all non-Gmail connectors — 895 failed jobs, zero data ingested across Drive, Docs, Discord, YouTube, Excel, OneDrive, Gumroad

User rome (jerome-v:crawford) reports a systematic memory ingestion failure across all connectors except Gmail. Well-structured bug report with clear reproduction steps. Failure summary Gmail: Working — 45 nodes ingested successfully. All other connectors: BROKEN — 895 failed jobs total, zero data ingested. Affected connectors: Google Drive, Google Docs, Discord, YouTube, Excel, OneDrive, Gumroad. Additional behaviors observed Manual ingest_document call reports success, but the Memory Tree cannot read the data back — suggesting the ingestion write path and the read/query path are out of sync, or data is being written to the wrong location. Reconnecting services did not trigger a fresh pipeline run — the failure state persists across reconnects with no retry or reset behavior. The Gmail-only success strongly suggests a connector-specific processing path difference, not a general pipeline outage. The issue may be in how non-Gmail connectors format or route their data into the ingestion queue. Also consistent with the previous ticket (165 failed jobs, Gmail idle with 155 pending) — this may be a broader pipeline regression affecting multiple users.