test: migrate stats/incr/nanmmda to ULP-based assertions

[kgryte]

Resolves a part of #11352.

Description

What is the purpose of this pull request?

This pull request:

• migrates the tests for stats/incr/nanmmda from a computed relative-tolerance comparison (delta = abs( expected - actual ) / tol = 1.0 * EPSILON * abs( expected ), asserted via t.equal( delta <= tol, ... )) to a ULP-difference assertion using @stdlib/assert/is-almost-same-value.
• applies the migration to the single tolerance-based assertion site in test/test.js. The package has no test/test.native.js, and the remaining assertions in the file are exact comparisons against null and exactly representable values, which are correct as-is and are left unchanged.
• removes the now-unused @stdlib/math/base/special/abs and @stdlib/constants/float64/eps requires, along with the delta and tol variable declarations.
• collapses the if ( actual === expected[i] ) { ... } else { ... } branch in the moving mean directional accuracy test into a single ULP assertion, matching the migrated idiom.

Only a test file is changed; no implementation, fixture, or documentation changes are included.

ULP bounds

Assertion	Previous tolerance	ULP bound
moving mean directional accuracy computed incrementally (11 values)	1.0 * EPSILON * abs( expected )	0

0 is the minimum integer bound N such that isAlmostSameValue( actual, expected[ i ], N ) holds at the assertion site, and it is minimal by construction, since N cannot be lowered further.

The bound was measured independently of the test harness by computing ulpDifference( actual, expected[ i ] ) directly at every element of the fixture, starting from a high bound (64, which passes) and tightening. The per-element ULP differences are 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0: the accumulator reproduces each hand-computed expected value (1, 1/2, 2/3, 1/3, 0) exactly. This is expected for this accumulator, since each returned value is a ratio of small integer counts over a fixed window, so no rounding error accumulates. Note that at N = 0, isAlmostSameValue reduces to the SameValue algorithm, which still distinguishes +0 from -0; the two zero-valued expectations are +0 on both sides, so the assertion holds.

make test TESTS_FILTER=".*/stats/incr/nanmmda/.*" was run three times at the final bound with identical results: 28/28 passing on every run, no failures, ruling out FMA/architecture-dependent flakiness on this platform. make lint-javascript-tests TESTS_FILTER=".*/stats/incr/nanmmda/.*" is clean.

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• [RFC]: Migrate math/base/special packages from relative tolerance testing to ULP difference testing (tracking issue) #11352

Questions

Any questions for reviewers of this pull request?

The pre-existing test already had an exact-equality fast path (if ( actual === expected[i] )) with the relative-tolerance comparison only as a fallback, and the fallback turns out never to be taken. The migration therefore collapses to N = 0, matching the idiom used elsewhere in the migration (a bound of 0 is used at 829 assertion sites already on develop). Flagging in case a reviewer would prefer a plain t.strictEqual( actual, expected[ i ], ... ) at this site instead; no such change was made, since it would go beyond migrating the assertion.

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

The resulting diff mirrors the already-merged migration for the sibling package stats/incr/mskewness (#15101), which has the same accumulator test structure.

Two environment notes, neither of which affected verification:

• make install-node-modules failed with ETARGET for es-object-atoms@^1.1.2. In this sandbox the failure was not a stale cache: npm view es-object-atoms versions reports 1.1.1 as the highest published version, so the constraint is unsatisfiable against the live registry. It was worked around by temporarily adding an overrides entry pinning es-object-atoms to 1.1.1 in the root package.json, running npm install, and then restoring package.json. The override is local only and is not part of this diff; make init and the full toolchain then worked normally.
• The lint-editorconfig-files pre-commit step could not download the editorconfig-checker binary in this sandbox and was skipped via SKIP_LINT_EDITORCONFIG; the changed lines were checked manually for LF line endings, tab indentation, no trailing whitespace, and a final newline. All other pre-commit lint steps ran normally.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of AI assistance?

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

This PR was written primarily by Claude Code, running as an unattended scheduled task. It selected the package, studied previously migrated packages in the same family to match the established idiom, performed the migration, and determined the minimum passing ULP bound empirically at the assertion site.

---

@stdlib-js/reviewers

🤖 Generated with Claude Code

https://claude.ai/code/session_01Q9twLhajFJVu4yHjgYdCzM

---

Generated by Claude Code

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
stats/incr/nanmmda	$\\color{green}152/152$ $\\color{green}+100.00\\%$	$\\color{green}8/8$ $\\color{green}+100.00\\%$	$\\color{green}2/2$ $\\color{green}+100.00\\%$	$\\color{green}152/152$ $\\color{green}+100.00\\%$

The above coverage report was generated for the changes in this PR.