feat(keywords): add bounded market insights core

[hubab1]

Summary

• add one shared read-only service for explicit app, storefront, platform, and optional keyword scopes
• distinguish ranked, not ranked, never refreshed, untracked, failed with cached evidence, failed without evidence, and bounded-scan unavailable states
• aggregate best, worst, average, spread, freshness, partial-data reasons, and stored estimated-difficulty provenance without network work
• use fixed-size scope-bound cursors, bounded track/snapshot/status scans, generation-safe status resolution, cancellation checks, and deterministic ordering

Safety and review status

• Read-only: no provider calls, persistence writes, or automatic all-market fan-out
• Stacked on PR feat(keywords): expose estimated difficulty evidence #39 because it consumes the reviewed ranking, status, and estimated-difficulty persistence stack
• Kept as a draft/category-B review until ranking-source and estimated-difficulty semantics are accepted
• The app was built and code-sign verified without launching it against the live V4 store

Verification

• 20/20 focused KeywordMarketInsightsService tests
• 424/424 full repository tests on the isolated core commit
• signed Debug build and strict codesign verification
• final independent semantic review: P1/P2 clear

Attribution

Independent rewrite informed by the useful market-comparison idea in upstream WIP commit akshaynexus@441b871. The unsafe global fan-out implementation was not carried forward. Source link only; no co-author trailer.