feat(providers): refresh available Codex models, add refresh button, and add Astra pricing

[atomantic]

Summary

• Refreshes available models for Codex CLI and Codex TUI to include GPT-6 Astra, GPT-5.6 Sol/Terra/Luna, GPT-5.5, GPT-5.4, GPT-5.4 Mini, and GPT-5.3 Codex Spark.
• Implements _fetchCodexModels via codex app-server stdio JSON-RPC (initialize -> model/list) and wires Codex in MODEL_FETCHERS to expose canRefreshModels: true, lighting up the "Refresh Models" button on Codex provider cards.
• Preserves prior catalogs in PRIOR_CODEX_MODEL_CATALOGS for safe auto-migration of existing installs.
• Updates shipped defaults in data.reference/providers.json and server/lib/aiToolkit/defaults/providers.sample.json, as well as the active instance.
• Adds GPT-6 Astra pricing data ($10/1M in, $50/1M out, $1/1M cache read, $12.50/1M cache write) in server/lib/modelPricing.js for usage estimates.
• Adds comprehensive unit and integration tests across client and server.

Test Plan

• server/lib/aiToolkit/providers.test.js: Verified auto-migration for sentinel-only and prior catalogs, as well as refreshProviderModels with codex app-server.
• server/lib/aiToolkit/internal/codex.test.js & server/lib/aiToolkit/internal/modelFetchers.test.js: Verified helper and model fetcher mappings.
• server/lib/modelPricing.test.js: Verified GPT-6 Astra exact and family resolution rates.
• client/src/components/providers/ProviderCard.test.jsx: Verified "Refresh Models" button renders on Codex card.
• Full server test suite passes (npm test in server).