fix: address 3 bugs found during PR review

fix: prevent false positive issues with universal suffix stripping and reverse alias checking

fix: add verified xAI Grok 4.20 models and fix updater variant handling (Issues #28, #29)

fix: prevent auto-updater from re-flagging deprecated models as NEW

fix: resolve auto-updater false positives for gpt-5.4-pro and gemini-3-pro

docs: update README for GPT-5.4 and GPT-5.4 Pro launch

feat: harden auto-updater with dedup, circuit breaker, normalization, and API-first fetching

fix: correct GPT-5.4 specs and add GPT-5.4 Pro

fix: deprecate 4 removed models and add 9 new models

docs: update README with accurate counts and smoother onboarding

fix: xAI regex over-matches docs page paths as model IDs

fix: updater false positives from Mistral naming and missing alias suffixes

fix: GPT-4.1 context window typo and stale model count in docs

fix: 3 updater bugs - saba exclude, substring matching, slice aliasing

fix: 7 bugs across tools, middleware, and server