fix: use proper chat template for warmup prompt

The warmup was using an unformatted plain text prompt, causing the processor
to generate 0 tokens and leading to 'tokens: 0, features 256' error.

Updated performWarmup() to format the prompt using the same chat template
and system prompt as normal inference, ensuring proper tokenization.

Fixes the warmup failure error that appeared after refactoring.

Author: deepdevjose

.prettierrc

@@ -0,0 +1,11 @@
+{
+  "semi": true,
+  "singleQuote": true,
+  "trailingComma": "es5",
+  "tabWidth": 4,
+  "useTabs": false,
+  "printWidth": 100,
+  "bracketSpacing": true,
+  "arrowParens": "always",
+  "endOfLine": "lf"
+}