id(ssh): fix #211 reconciliation gaps: total records cap, document output through send, per-world document store #548

What and why

The reports flagged three gaps in #211: /records drains all pages with only a per-page 384 KiB cap; document handlers write to the channel inside recv rather than through send; and documents read the server's local blob store, not a per-world store. These were flagged while reconciling #211 and were not addressed before it merged.

Source

• RPT §4 'Known gaps to fix while reconciling'
• RPT §7 Step 1

RPT = doc/2026-10-18T18-40-00Z_report_platform_world_state/2026-10-18T18-40-00Z_report_platform_world_state.md, REF = doc/2026-10-18T00-00-00Z_reference_platform_state/2026-10-18T00-00-00Z_reference_platform_state.md. Both are uncommitted reports in the platform-world-apps-migration worktree.

Current status

Partial. pkgs/id/src/world_ssh.rs:595 MAX_RECORDS_TERMINAL_BYTES = 384 * 1024, enforced while rendering (:792). Whether that bounds the whole drain or each page needs a test (not verified). Documents are still read in the recv command loop from the server blob store: list_public_documents(&self.blobs, ...) at :455. No per-world store.

Acceptance criteria

• A test shows whether MAX_RECORDS_TERMINAL_BYTES bounds the whole /records drain; if it only bounds a page, a total cap is added
• Document handlers write through send, not from inside recv
• Documents read a per-world store rather than the server's local blob store, or a recorded decision explains why not

Size

S

Prerequisites

None.