FIX DUMMY: avoid spurious out-of-range threshold warning

[JayeshSuryavanshi]

Submissions Basics:

• Followed the guidelines in the Contributing document.
• Checked there are no other open Pull Requests for the same change (0 open PRs).
• Tied the PR to a specific issue: Fixes DUMMY thresholder emits a spurious 'threshold outside the range' warning on every default call #16.

Type of Change:

• Bug fix

All Submissions Cores:

• Added an explanation of what the changes do and why (below).
• Written new unit tests for the change.
• All new and existing unit tests pass locally.

What & why

DUMMY emits a spurious UserWarning ("Computed threshold 1.0000000000000002 is outside the range of 0 and 1") on every default call — the repo's own suite triggers it 10 times, all from tests/test_dummy.py.

Decision scores are min-max normalized to [0, 1], so np.percentile(decision, perc) is in [0, 1]. eval adds eps (limit = percentile + eps) so the top-scoring point stays an inlier at contam=0 (cut uses decision >= limit). It then passed that eps-bumped limit to _check_threshold, so a maximum percentile (== 1) made limit == 1 + eps > 1 and tripped the range guard — even though the threshold is valid and the labels are correct.

This validates the in-range percentile with _check_threshold instead of the eps-bumped limit. The stored thresh_ and the returned labels are unchanged (all pre-existing DUMMY tests still pass); only the spurious warning is gone.

• Added tests/test_dummy.py::test_no_spurious_threshold_warning, asserting no out-of-range warning across the contam grid.
• Added a CHANGES.txt entry (happy to retag the version if you'd prefer a different one).

ruff check clean; pytest tests/test_dummy.py → 20 passed, 0 warnings.

[JayeshSuryavanshi]

Hi @KulikDM — this fixes #16 (the spurious out-of-range warning DUMMY emits on every default call); small and test-only, with a regression test. Ready for review whenever you have a moment — thanks!