Refresh deepseek models. (#4660)

### What problem does this PR solve?


### Type of change

- [x] New Feature (non-breaking change which adds functionality)

conf/llm_factories.json

                 {
                     "llm_name": "deepseek-chat",
                     "tags": "LLM,CHAT,",
-                    "max_tokens": 32768,
+                    "max_tokens": 64000,
                     "model_type": "chat"
                 },
                 {
-                    "llm_name": "deepseek-coder",
+                    "llm_name": "deepseek-reasoner",
                     "tags": "LLM,CHAT,",
-                    "max_tokens": 16385,
+                    "max_tokens": 64000,
                     "model_type": "chat"
                 }
             ]