Update LLM support for llama.cpp and enhance configuration management

- Added support for `llamacpp-models.yaml` in `.gitignore` and implemented logic to copy it in `gpu-rent.ps1` and `gpu-rent.sh`.
- Enhanced CLI to prompt for llama.cpp model presets during setup and execution, improving user experience.
- Updated configuration handling to include `llamacpp_models_manifest` and related functions for managing llama.cpp models.
- Improved documentation in `cli.md` and `llm.md` to reflect changes in llama.cpp integration and model management.
- Refactored provisioning logic to handle llama.cpp model downloads and configurations effectively.

tests/test_llamacpp_models.py

@@ -0,0 +1,31 @@
+from pathlib import Path
+
+from gpu_rent.llm_runtime import (
+    gguf_filename_from_url,
+    parse_llamacpp_models,
+    write_llamacpp_models_preset,
+)
+
+
+def test_gguf_filename_from_url():
+    url = (
+        "https://huggingface.co/org/repo/resolve/main/"
+        "Qwen2.5-3B-Instruct-Q4_K_M.gguf"
+    )
+    assert gguf_filename_from_url(url) == "Qwen2.5-3B-Instruct-Q4_K_M.gguf"
+
+
+def test_write_and_parse_llamacpp_preset(tmp_path: Path):
+    path = tmp_path / "llamacpp-models.yaml"
+    write_llamacpp_models_preset(path, "light")
+    entries = parse_llamacpp_models(path)
+    assert len(entries) == 1
+    assert entries[0].default is True
+    assert "Qwen2.5-3B" in entries[0].url
+    assert entries[0].url.startswith("https://")
+
+
+def test_parse_llamacpp_empty(tmp_path: Path):
+    path = tmp_path / "llamacpp-models.yaml"
+    write_llamacpp_models_preset(path, "empty")
+    assert parse_llamacpp_models(path) == []

tests/test_server_plan.py

@@ -0,0 +1,61 @@
+from gpu_rent.inventory import FlavorInfo
+from gpu_rent.ux import prompt_server_plan
+from gpu_rent.varsfile import parse_vars_file, upsert_vars
+
+
+def _cfg_stub(monkeypatch, data_gb: int = 100):
+    monkeypatch.setenv("OS_AUTH_URL", "https://example.invalid/identity/v3")
+    monkeypatch.setenv("OS_USER_DOMAIN_NAME", "999")
+    monkeypatch.setenv("OS_USERNAME", "svc")
+    monkeypatch.setenv("OS_PASSWORD", "secret")
+    monkeypatch.setenv("OS_PROJECT_ID", "proj")
+    monkeypatch.setenv("OS_REGION_NAME", "ru-7")
+    monkeypatch.setenv("GPU_RENT_AZ", "ru-7a")
+    monkeypatch.setenv("DATA_VOLUME_SIZE_GB", str(data_gb))
+    from gpu_rent.config import load_config
+
+    return load_config(require_auth=True)
+
+
+def test_prompt_server_plan_with_ranked(monkeypatch, tmp_path):
+    cfg = _cfg_stub(monkeypatch)
+    monkeypatch.setattr("gpu_rent.paths.app_root", lambda: tmp_path)
+    f1 = FlavorInfo(
+        id="a", name="small", vcpus=4, ram_mb=16384, disabled=False, extra={}, label="4090-24"
+    )
+    f2 = FlavorInfo(
+        id="b", name="big", vcpus=12, ram_mb=65536, disabled=False, extra={}, label="4090-48"
+    )
+
+    monkeypatch.setattr(
+        "gpu_rent.ux.list_ranked_flavors",
+        lambda flavors, cfg: [f1, f2],
+    )
+    answers = iter(["2", "200", "n"])
+
+    def ask(msg: str, default: str = "") -> str:
+        return next(answers)
+
+    remembered: list[bool] = []
+
+    plan = prompt_server_plan(
+        cfg,
+        ["x"],
+        picked=f1,
+        spot=True,
+        ask=ask,
+        confirm=lambda m: remembered.append(True) or False,
+    )
+    assert plan.flavor.id == "b"
+    assert plan.data_gb == 200
+    assert plan.spot is False
+    assert remembered
+
+
+def test_upsert_vars(tmp_path):
+    path = tmp_path / "gpu-rent.vars"
+    path.write_text("# c\nLLM_RUNTIME=none\n", encoding="utf-8")
+    upsert_vars(path, {"LLM_RUNTIME": "llamacpp", "DATA_VOLUME_SIZE_GB": "200"})
+    data = parse_vars_file(path)
+    assert data["LLM_RUNTIME"] == "llamacpp"
+    assert data["DATA_VOLUME_SIZE_GB"] == "200"