ci: fix reviewer models — sonnet uses Anthropic, gpt uses GPT-5

The matrix was wrong: "sonnet" was running GPT-5 and "gpt" was running
GPT-4.1. Now:
- sonnet → Claude Sonnet 4.6 via HAI Anthropic endpoint
- gpt → GPT-5 via HAI OpenAI endpoint
- security → GPT-5 via HAI OpenAI endpoint

Each matrix entry specifies its own provider and base_url.

.gitea/workflows/ci.yml

@@ -36,21 +36,6 @@ jobs:
             provider: openai
             base_url: http://100.86.77.84:6655/openai/v1
             model: gpt-5
-          - name: gpt41
-            token_secret: SONNET_REVIEW_TOKEN
-            provider: openai
-            base_url: http://100.86.77.84:6655/openai/v1
-            model: gpt-4.1
-          - name: gpt5-mini
-            token_secret: SONNET_REVIEW_TOKEN
-            provider: openai
-            base_url: http://100.86.77.84:6655/openai/v1
-            model: gpt-5-mini
-          - name: gpt41-mini
-            token_secret: SONNET_REVIEW_TOKEN
-            provider: openai
-            base_url: http://100.86.77.84:6655/openai/v1
-            model: gpt-4.1-mini
           - name: security
             token_secret: SECURITY_REVIEW_TOKEN
             provider: openai