chore: update keywords

[Planeshifter]

Follow-up fixes for commits merged to develop between 2026-09-27 14:57 -0700 (e1724483) and 2026-09-28 03:38 -0700 (41f72749).

Description

What is the purpose of this pull request?

This pull request:

• In 3dc8a4f, blas/ext/base/ndarray/gfill-less-than/package.json used "less than" and a stray "lt" in keywords; neither matches the base package or the gfill-greater-than sibling. Collapsed both into "less-than" to match convention.

Review window: 38 first-parent commits, 420 files. The batch breaks down as follows, and everything below was reviewed and found clean:

• C addon backfill (14 commits) — blas/ext/base/ndarray/{dany,sany,dnone,snone,dxsa,sxsa,daxpb,saxpb,dwxsa,done-to,dzero-to} and stats/base/ndarray/{smaxabssorted,dminsorted,smediansorted}. Precision tokens (double/Float64Array vs float/Float32Array), header guards, exported symbol names, N-API argument counts, manifest.json deps and binding.gyp/include.gypi/src/Makefile all match their own package; no d↔s crossover.
• New packages (6 commits) — blas/ext/base/gpad-edge, blas/ext/base/dleft-pad-edge, blas/ext/base/gindex-of-greater-than-sorted, blas/ext/base/ndarray/gfill-less-than, math/base/special/cothf, fft/base/fftpack/ndarray/float64/sinqi. Every documented numeric example was recomputed by hand, and the pad/binary-search implementations were brute-forced against independent references across negative, zero and non-unit strides plus view offsets.
• Namespace wiring (3 commits) — all 14 newly registered symbols resolve, target packages exist, insertion points alphabetical.
• ULP test migrations (6 commits) — argument order, float32 vs float64 helper choice, and ULP counts verified against the real fixtures, including the two tightenings (1.0*EPS → 0 ULP in stats/base/dists/logistic/mode, 2.0*EPS → 0 ULP in stats/base/dists/binomial/variance).
• Docs, chore and CI (9 commits) — including the jq null-guard in process_metadata.yml (d7c4ea0) and the missing-test-directory skip in run_tests_coverage/run (1e56ff8).

Related Issues

Does this pull request have any related issues?

No.

Questions

Any questions for reviewers of this pull request?

No.

Other

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

Validation. Four independent reviewers ran over the window: two style/convention audits comparing each changed or new package against an established reference package of comparable complexity and checking it against docs/style-guides, and two bug scans (one restricted to the diff text, one permitted to read callees in order to reason about the introduced logic). Coverage included precision-token and sibling cross-comparison across the C addon backfill, hand-recomputation of every documented numeric example, brute-force differential testing of the new pad and binary-search routines over negative/zero/non-unit strides and view offsets, C memory safety and build wiring, the ULP migrations, and the two CI/shell changes.

Findings were merged and de-duplicated, then anything that could not be independently re-verified by re-reading the diff, lacked a concrete fix, or fell outside the window was dropped. One issue survived. The "less than"/"lt" keywords are the only occurrences of either string in any package.json in the repo; "less-than" is used by three others, including the base package this one wraps.

Deliberately excluded: anything requiring interpretation or a judgment call. Two such items were raised and dropped — smaxabssorted/benchmark/benchmark.native.js naming its import isnan where smediansorted uses isnanf (identical behavior), and smaxabssorted doc comments omitting "sorted" where its siblings include it. Also out of scope: the stats/strided/* callees returning X[0] while ignoring offsetX when N == 1 || strideX == 0, which is pre-existing and untouched by this window.

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

If you answered "yes" to using AI assistance, please provide a short disclosure indicating how you used AI assistance. This helps reviewers determine how much scrutiny to apply when reviewing your contribution. Example disclosures: "This PR was written primarily by Claude Code." or "I consulted ChatGPT to understand the codebase, but the proposed changes were fully authored manually by myself.".

This PR was produced by Claude Code running as a scheduled review of commits merged to develop in the preceding 24 hours. The review, the filtering of findings, and the one-line fix were all authored by Claude Code. Opened as a draft for maintainer audit before promotion.

---

🤖 Generated with Claude Code

https://claude.ai/code/session_01KiNTzHsWDjMc7CagsFEcpz

---

Generated by Claude Code

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
blas/ext/base/ndarray/gfill-less-than	$\\color{green}167/167$ $\\color{green}+100.00\\%$	$\\color{green}5/5$ $\\color{green}+100.00\\%$	$\\color{green}1/1$ $\\color{green}+100.00\\%$	$\\color{green}167/167$ $\\color{green}+100.00\\%$

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