Enhance LLM runtime and installation script for Linux support

- Added a new function `pick_llamacpp_linux_asset_url` to select appropriate Linux release assets, prioritizing Ubuntu CUDA and Vulkan options while excluding Windows and macOS binaries.
- Updated the installation script to build `llama-server` from source when Linux CUDA binaries are unavailable, improving compatibility and flexibility.
- Revised documentation to reflect changes in asset handling and installation procedures.
- Added tests to validate the new asset selection logic, ensuring correct behavior in various scenarios.

tests/test_llm_runtime.py

@@ -7,6 +7,7 @@ from gpu_rent.llm_runtime import (
     normalize_runtime,
     parse_llamacpp_models,
     parse_ollama_models,
+    pick_llamacpp_linux_asset_url,
     remap_llamacpp_url,
     write_ollama_models_preset,
 )
@@ -86,3 +87,39 @@ def test_remap_dead_bartowski_abliterate_url(tmp_path: Path):
     entries = parse_llamacpp_models(path)
     assert len(entries) == 1
     assert entries[0].url == fixed
+
+
+def test_pick_llamacpp_linux_asset_skips_windows_cuda():
+    assets = [
+        {
+            "name": "cudart-llama-bin-win-cuda-12.4-x64.zip",
+            "browser_download_url": "https://example/cudart-win.zip",
+        },
+        {
+            "name": "llama-b10545-bin-win-cuda-12.4-x64.zip",
+            "browser_download_url": "https://example/win-cuda.zip",
+        },
+        {
+            "name": "llama-b10545-bin-ubuntu-x64.tar.gz",
+            "browser_download_url": "https://example/ubuntu-cpu.tar.gz",
+        },
+        {
+            "name": "llama-b10545-bin-ubuntu-vulkan-x64.tar.gz",
+            "browser_download_url": "https://example/ubuntu-vulkan.tar.gz",
+        },
+    ]
+    assert pick_llamacpp_linux_asset_url(assets) == "https://example/ubuntu-vulkan.tar.gz"
+
+
+def test_pick_llamacpp_linux_asset_prefers_ubuntu_cuda():
+    assets = [
+        {
+            "name": "llama-b1-bin-ubuntu-vulkan-x64.tar.gz",
+            "browser_download_url": "https://example/vulkan.tar.gz",
+        },
+        {
+            "name": "llama-b1-bin-ubuntu-cuda-12.4-x64.tar.gz",
+            "browser_download_url": "https://example/cuda.tar.gz",
+        },
+    ]
+    assert pick_llamacpp_linux_asset_url(assets) == "https://example/cuda.tar.gz"