Reduce measured code-health debt

[sarthakagrawal927]

Follow-up to #32. The adoption change establishes no-regression gates; this issue owns the measured legacy debt and the work required to raise Knowledge Base toward Fleet targets.

Current baseline (2026-08-12):

• Format: 86 active JavaScript/TypeScript files remain outside the Biome format baseline.
• Complexity: 57 threshold violations across 1,578 production functions and 20,953 function NLOC; maximum CCN 126, function length 6,151, and 9 parameters.
• Duplication: 60 clone groups and 712 duplicated lines (3.0791%) across 58 production files.
• Worker coverage: 56.36% lines, 48.24% branches, 56.26% functions, and 53.99% statements across 360 passing tests and 1 skipped test.
• Dashboard coverage: 100% lines/statements/functions and 97.05% branches across the explicitly gated project-scope and formatting helpers; the route and page layer remains outside this initial five-test gate.
• Unused surface: 4 exports and 7 types are reviewed and regression-gated; high-confidence unused files/dependencies are zero after removing two unused dashboard packages.
• Dependency risk: zero critical advisories; high advisories are root docs 14, dashboard 11, Worker 5, and landing 3. New advisory IDs fail the gate.
• Cycles and inline suppressions: zero.

Smallest next actions:

• Add dashboard route, Access boundary, and Pages Function tests before widening its coverage include set.
• Split the Worker index and parser hot spots only behind focused tests.
• Consolidate clone groups only where retrieval, migration, and protocol semantics are genuinely shared.
• Format active code in bounded package batches and lower the exact file ceiling after each batch.
• Verify framework and public-client boundaries before removing reviewed exports or types.
• Upgrade or override vulnerable transitive chains without an unsafe docs or build-tool major migration.

Review the accepted baselines by 2026-11-12.