Add GPT-6.1 Sol and zero-cost Auto-review pricing (v2.7.2)

[zJay26]

Changes

GPT-6.1 Sol usage currently remains unpriced, and separate Auto-review usage has no default rate. Add the published GPT-6.1 Sol input/cache/write/output rates and Codex Fast weighting, and default the exact codex-auto-review model to zero cost while retaining raw token counts and pricing coverage.

Auto-review zero pricing is explicitly documented as a local accounting policy based on the maintainer-reported Tibo statement, not a verified published API price. Existing manual overrides remain valid and take precedence; resetting one restores zero pricing and refreshes aggregate/session estimates. Main-model usage and unknown models retain their existing semantics.

Update the Dashboard, demo, bilingual docs, and v2.7.2 version/release notes. Fix the timezone assumptions in two existing date tests. No schema change or historical rebuild is required.

Validation

• Full Go tests and go vet passed locally.
• All 27 Dashboard/demo browser tests passed, including GPT-6.1 Sol snapshots, Fast estimates, zero-cost coverage, and resetting an older Auto-review override.
• Windows build, JavaScript syntax checks, and git diff --check passed.
• Cross-platform CI and release artifact verification will run before delivery.