feat: GPU detection for LLM server (NVIDIA/AMD/Intel/Vulkan + CPU fallback)

- Added start-llama-server.bat with cross-brand GPU auto-detection:
  * Detection 1: nvidia-smi for NVIDIA
  * Detection 2: WMIC for AMD, Intel Arc, Intel HD/Iris/UHD
  * Detection 3: vulkaninfo as final fallback probe
  * Falls back to CPU-only build when no GPU found
- Downloaded llama-server b9947 builds:
  * Vulkan build (91 MB) - supports all GPU brands
  * CPU build (45 MB) - fallback in cpu/ subdir
- Downloaded Qwen2.5-Coder-1.5B Q4_K_M model (1.1 GB)
- Fixed run-hermes.bat polling: replaces fixed 5s timeout with
  curl-based readiness polling (up to 30 attempts, 1s apart)
- All .bat files verified: ASCII text, CRLF line terminators
- Added .gitignore for build artifacts and large binaries

build/output/start-llama-server.bat

@@ -0,0 +1,125 @@
+@echo off
+:: ==============================================================================
+:: Hermes Portable Rescue - llama-server launcher with GPU detection
+:: ==============================================================================
+:: Detects GPU brand (NVIDIA, AMD, Intel, integrated) and launches the
+:: appropriate llama-server build (Vulkan for GPU, CPU-only fallback).
+:: ==============================================================================
+setlocal enabledelayedexpansion
+
+cd /d "%~dp0"
+
+set LLAMA_DIR=%~dp0bin\llama
+set MODEL_DIR=%~dp0bin\ollama
+set MODEL=%MODEL_DIR%\qwen2.5-coder-1.5b-q4_k_m.gguf
+set PORT=11434
+set GPU_FOUND=0
+
+echo.
+echo ============================================
+echo   Hermes LLM Server - GPU Detection
+echo ============================================
+echo.
+
+:: --- Detection 1: nvidia-smi (NVIDIA) --------------------------------------
+nvidia-smi >nul 2>&1
+if !errorlevel! equ 0 (
+    for /f "tokens=*" %%a in ('nvidia-smi --query-gpu=name --format=csv,noheader 2^>nul') do (
+        set GPU_NAME=%%a
+    )
+    echo [OK] NVIDIA GPU detected: !GPU_NAME!
+    set GPU_FOUND=1
+)
+
+::  Detection 2: WMIC - check for AMD/Intel GPUs 
+if !GPU_FOUND! equ 0 (
+    wmic path win32_videocontroller get name 2>nul | findstr /i "nvidia" >nul
+    if !errorlevel! equ 0 (
+        echo [OK] NVIDIA GPU detected via WMI
+        set GPU_FOUND=1
+    ) else (
+        wmic path win32_videocontroller get name 2>nul | findstr /i "amd radeon" >nul
+        if !errorlevel! equ 0 (
+            echo [OK] AMD GPU detected via WMI
+            set GPU_FOUND=1
+        ) else (
+            wmic path win32_videocontroller get name 2>nul | findstr /i "intel arc" >nul
+            if !errorlevel! equ 0 (
+                echo [OK] Intel GPU detected via WMI
+                set GPU_FOUND=1
+            ) else (
+                wmic path win32_videocontroller get name 2>nul | findstr /i "intel hd intel iris intel uhd" >nul
+                if !errorlevel! equ 0 (
+                    echo [OK] Intel integrated graphics detected via WMI
+                    set GPU_FOUND=1
+                )
+            )
+        )
+    )
+)
+
+::  Detection 3: vulkaninfo (direct Vulkan probe) 
+if !GPU_FOUND! equ 0 (
+    where vulkaninfo >nul 2>&1
+    if !errorlevel! equ 0 (
+        vulkaninfo --summary 2>nul | findstr /c:"Vulkan Instance" >nul
+        if !errorlevel! equ 0 (
+            echo [OK] Vulkan-capable device detected
+            set GPU_FOUND=1
+        )
+    )
+)
+
+::  Check model exists 
+if not exist "!MODEL!" (
+    echo [ERROR] Model not found: !MODEL!
+    echo.
+    echo Please download a GGUF model and place it at:
+    echo   !MODEL!
+    echo.
+    echo Recommended: Qwen 2.5 Coder 1.5B Q4_K_M
+    echo   https://huggingface.co/Qwen/Qwen2.5-Coder-1.5B-Instruct-GGUF
+    echo.
+    pause
+    exit /b 1
+)
+
+::  Launch 
+if !GPU_FOUND! equ 1 (
+    echo.
+    echo [*] Starting llama-server with GPU acceleration (Vulkan)...
+    echo [*] Model: qwen2.5-coder-1.5b-q4_k_m.gguf
+    echo [*] Server: http://127.0.0.1:!PORT!/v1
+    echo [*] Flags: --n-gpu-layers 99 --mlock --cont-batching -c 4096
+    echo.
+    echo [*] Press Ctrl+C to stop the server.
+    echo.
+
+    "!LLAMA_DIR!\llama-server.exe" ^
+        --host 127.0.0.1 --port !PORT! ^
+        -m "!MODEL!" ^
+        --n-gpu-layers 99 ^
+        --mlock ^
+        --cont-batching ^
+        -c 4096
+) else (
+    echo.
+    echo [*] No compatible GPU found. Starting CPU-only mode.
+    echo [*] Model: qwen2.5-coder-1.5b-q4_k_m.gguf
+    echo [*] Server: http://127.0.0.1:!PORT!/v1
+    echo [*] Flags: --mlock --cont-batching -c 4096
+    echo.
+    echo [*] Press Ctrl+C to stop the server.
+    echo.
+
+    "!LLAMA_DIR!\cpu\llama-server-cpu.exe" ^
+        --host 127.0.0.1 --port !PORT! ^
+        -m "!MODEL!" ^
+        --mlock ^
+        --cont-batching ^
+        -c 4096
+)
+
+echo.
+echo [*] Server exited with code !errorlevel!
+pause