feat(benchmark): support multi-dataset evaluation with configurable top-k list

- Evaluate multiple datasets in a single run (CIFAR10 + CIFAR100)
- Report Recall@K for a list of K values from one underlying search
- Save results to disk: summary.md, metrics.csv, per-dataset predictions.csv, confusion_matrix.csv
- Richer evaluation output: query_labels, topk_ids, topk_labels for downstream analysis
- Add --dataset and --top-k CLI overrides for benchmark command
- Update config schema: dataset→datasets, top_k→top_k_list

mini-nav/commands/benchmark.py

@@ -1,3 +1,5 @@
+"""Benchmark CLI command."""
+
 import typer
 from commands import app
 
@@ -8,19 +10,59 @@ def benchmark(
     model_path: str | None = typer.Option(
         None, "--model", "-m", help="Path to compressor model weights"
     ),
+    dataset: list[str] | None = typer.Option(
+        None,
+        "--dataset",
+        "-d",
+        help=(
+            "Override datasets (HuggingFace path, repeatable). "
+            "Each value creates a single-entry dataset list."
+        ),
+    ),
+    top_k: list[int] | None = typer.Option(
+        None,
+        "--top-k",
+        "-k",
+        help="Override top-K values (repeatable). E.g. --top-k 1 --top-k 5 --top-k 10",
+    ),
 ):
+    """Run benchmark evaluation across configured datasets.
+
+    Without flags, runs all datasets and K values from config.yaml.
+    Use --dataset to override datasets, --top-k to override K values.
+    """
     from benchmarks import run_benchmark
     from configs import cfg_manager
+    from configs.models import DatasetSourceConfig
 
     config = cfg_manager.get()
     benchmark_cfg = config.benchmark
+
     if model_path:
         config.model.compressor_path = model_path
 
-    _ = run_benchmark(
+    # CLI overrides
+    if dataset:
+        benchmark_cfg.datasets = [
+            DatasetSourceConfig(
+                source_type="huggingface",
+                path=d,
+            )
+            for d in dataset
+        ]
+
+    if top_k:
+        benchmark_cfg.task.top_k_list = sorted(set(top_k))
+
+    model_name = "hash_compressor" if config.model.compressor_path else "dinov2"
+    output_root = config.output.directory / "benchmark"
+
+    results = run_benchmark(
         model=None,
         processor=None,
         config=benchmark_cfg,
         model_config=config.model,
-        model_name="hash_compressor" if config.model.compressor_path else "dinov2",
+        model_name=model_name,
+        output_root=output_root,
     )
+    return results