Enhance CLI options and idle-killer functionality

- Added a `--update` option to the `up` command in the CLI for forced updates during warm ACTIVE states.
- Improved the idle-killer logic with new functions to check for reusable credentials and refresh settings on existing remote credentials.
- Enhanced the `seed_civitai` function to implement caching for model jobs, optimizing the provisioning process.
- Updated the installation script for Ollama to prevent unnecessary restarts when the service configuration has not changed.
- Added diagnostics for model job caching and improved error handling in various functions.

tests/test_warm_up_skips.py

@@ -0,0 +1,96 @@
+"""Warm re-up shortcuts: no git cascade, killer reuse, seed/ollama skips."""
+
+from pathlib import Path
+
+from gpu_rent.idle_killer import _remote_killer_reusable
+from gpu_rent.provision import _models_manifest_fp, seed_civitai
+
+
+def test_models_manifest_fp_stable(tmp_path):
+    man = tmp_path / "models.yaml"
+    man.write_text("models:\n  - kind: checkpoint\n", encoding="utf-8")
+
+    class Cfg:
+        models_manifest = man
+
+    a = _models_manifest_fp(Cfg())
+    b = _models_manifest_fp(Cfg())
+    assert a == b
+    assert len(a) == 16
+    man.write_text("models:\n  - kind: lora\n", encoding="utf-8")
+    assert _models_manifest_fp(Cfg()) != a
+
+
+def test_seed_civitai_skips_when_cache_dests_present(tmp_path, monkeypatch):
+    man = tmp_path / "models.yaml"
+    man.write_text("models:\n  - kind: checkpoint\n", encoding="utf-8")
+    from gpu_rent import provision
+
+    class Entry:
+        kind = "checkpoint"
+        url = "https://huggingface.co/org/model/resolve/main/a.safetensors"
+        version_id = None
+
+    monkeypatch.setattr(provision, "parse_models", lambda _p: [Entry()])
+
+    def fake_ssh(cfg, host, cmd, **kw):
+        if ".gpu-rent-models-jobs.json" in cmd and "cat" in cmd:
+            import json
+
+            return json.dumps(
+                {
+                    "fp": provision._models_manifest_fp(cfg),
+                    "dests": ["/mnt/swarm_data/Models/Stable-Diffusion/a.safetensors"],
+                }
+            )
+        if "ALL_OK" in cmd or "test -f" in cmd:
+            return "ALL_OK"
+        return ""
+
+    class Cfg:
+        models_manifest = man
+        civitai_api_token = "t"
+        civitai_api_host = "https://civitai.com"
+        hf_token = ""
+
+    logs = []
+    monkeypatch.setattr(provision, "run_ssh", fake_ssh)
+    seed_civitai(Cfg(), "1.2.3.4", logs.append)
+    assert any("skip" in x.lower() for x in logs)
+
+
+def test_remote_killer_reusable_parses_ok(monkeypatch):
+    class Cfg:
+        pass
+
+    def fake_ssh(cfg, host, cmd, **kw):
+        return (
+            "sid=abc-123\n"
+            "want=abc-123\n"
+            "creds=yes\n"
+            "armed=yes\n"
+            "rootcreds=yes\n"
+            "timer=active\n"
+        )
+
+    from gpu_rent import idle_killer
+
+    monkeypatch.setattr(idle_killer, "run_ssh", fake_ssh)
+    assert _remote_killer_reusable(Cfg(), "h", "abc-123") is True
+    assert _remote_killer_reusable(Cfg(), "h", "other") is False
+
+
+def test_install_ollama_skips_restart_when_unit_unchanged():
+    from importlib.resources import files
+
+    text = files("gpu_rent.remote").joinpath("install_ollama.sh").read_text(encoding="utf-8")
+    assert "cmp -s" in text
+    assert "skip restart" in text
+
+
+def test_cli_has_update_flag():
+    from gpu_rent import cli
+
+    src = Path(cli.__file__).read_text(encoding="utf-8")
+    assert '"--update"' in src
+    assert "force_update" in src