test: deduplicate suites and prune low-value cases

- extract shared helpers for dataset writers, scheduler construction, thread interleaving, hf roundtrips, and moe configs
- remove about 20 cases whose only assertions were format checks, restated declarations, fake-taxonomy duplicates, or test-local scaffolding
- strengthen weak cases into exact reference comparisons, positional mask checks, and deterministic outcomes
- replace two schedule factory smoke tests with cosine/sgdr formula assertions
- delete root-level CLI tests whose merge-priority facts are covered by tests/config/test_cli.py
- suite shrinks from 857 to 826 items; ruff format, import order, and pytest all green

tests/inference/test_sample.py

@@ -62,8 +62,9 @@ def test_top_p_nucleus_filtering():
     logits = torch.tensor([[10.0, 1.0, 1.0, 1.0, 1.0]])
     s = TopPStrategy(top_p=0.5)
     result = s.apply(logits.clone(), filter_value=-1e9)
+    # The dominant logit alone exceeds the nucleus mass; the rest are filtered.
     kept = (result > -1e9).sum().item()
-    assert kept >= 1
+    assert kept == 1
 
 
 def test_top_p_skip_when_one():