enhance: improve DeepSeek provider configuration in avante.nvim

Add timeout setting and restructure max_tokens into extra_request_body
for better API request handling and proper configuration structure.

lua/plugins/avante-nvim.lua

@@ -40,8 +40,13 @@ return {
           endpoint = "https://api.deepseek.com",
           -- AI model to use (specialized for programming)
           model = "deepseek-coder",
-          -- Maximum number of tokens in model response
+          timeout = 30000, -- Timeout in milliseconds
+          extra_request_body = {
+            temperature = 0.2,
             max_tokens = 8192,
+          },
+          -- Maximum number of tokens in model response
+          -- max_tokens = 8192,
           -- Tools disable flag (commented out)
           -- disable_tools = true,
         },