refactor(iqa): rename reserved-identifier surface + lint cascade sweep (ADR-0148)

[lusoris]

Summary

Rename every _iqa_* / struct _kernel / _ssim_int / _map_reduce / _map / _reduce / _context / _ms_ssim_* / _ssim_* / _alloc_buffers / _free_buffers symbol and the four underscore-prefixed header guards to non-reserved spellings across the IQA tree (21 files). Sweeps the ADR-0141 touched-file lint cascade that surfaced (~40 pre-existing warnings invisible to prior diff-based CI runs).

The rename's stated motivation in rebase-notes 0039 ("clear remaining reserved-identifier suppressions") turned out to be wrong on inspection — there were zero NOLINTs to clear. The actual win is the IQA tree is now baseline lint-clean, so future PRs touching ssim.c / ms_ssim.c / iqa/*.{c,h} / convolve_*.{c,h} no longer trip the touched-file lint cascade.

Bit-identical VMAF score: VMAF_CPU_MASK=0 vs =255 on Netflix golden pair src01_hrc00/01_576x324 with --feature float_ssim --feature float_ms_ssim → diff exit 0. Full vmaf_v0.6.1 score on frame 0: 83.856284 (unchanged from master).

Type

• refactor — no behavior change

Checklist

• Commits follow Conventional Commits.
• make format && make lint is green locally.
• Unit tests pass: meson test -C build → 32/32.
• Touched IQA / SIMD paths — bit-identical between scalar and SIMD on Netflix golden pair (diff exit 0).
• No new .c/.h/.cpp files added.
• Not a breaking change.

Netflix golden-data gate (ADR-0024)

• I did not modify any assertAlmostEqual(...) score in the Netflix golden Python tests.

Cross-backend numerical results

VMAF_CPU_MASK=0 vs =255 (scalar vs SIMD), --feature float_ssim --feature float_ms_ssim:
  All metrics  scalar-vs-simd = 0-ULP  (diff exit 0 on Netflix golden-pair #1)

Full vmaf_v0.6.1 model run, frame 0:
  vmaf="83.856284"  (identical pre/post this PR)

Deep-dive deliverables (ADR-0108)

• Research digest — no digest needed: mechanical rename + lint sweep.
• Decision matrix — captured in ADR-0148 §Alternatives considered.
• AGENTS.md invariant note — added to libvmaf/src/feature/AGENTS.md (IQA reserved-identifier rename + load-bearing NOLINT brackets).
• Reproducer / smoke-test command — pasted below under "Reproducer".
• CHANGELOG.md "lusoris fork" entry — bullet added under ### Changed in CHANGELOG.md.
• Rebase note — entry 0041 in docs/rebase-notes.md.

Reproducer

ninja -C build && meson test -C build

for mask in 0 255; do
  VMAF_CPU_MASK=$mask ./build/tools/vmaf \
    --reference python/test/resource/yuv/src01_hrc00_576x324.yuv \
    --distorted python/test/resource/yuv/src01_hrc01_576x324.yuv \
    --width 576 --height 324 --pixel_format 420 --bitdepth 8 \
    -m version=vmaf_v0.6.1 \
    --feature float_ssim --feature float_ms_ssim \
    -o /tmp/iqa_$mask.xml
done
diff <(grep -v fyi /tmp/iqa_0.xml) <(grep -v fyi /tmp/iqa_255.xml)
# expect exit 0 (bit-identical scalar vs SIMD on float_ssim / float_ms_ssim)

# Plus per-touched-file clang-tidy:
git diff --name-only HEAD | grep -E '\.(c|h)$' | grep -v arm64 | while read f; do
  clang-tidy -p build "$f" --quiet 2>&1 | grep -E "warning:|error:" | grep -v "Suppressed\|generated\|-warnings-as-errors"
done
# expect zero output

Known follow-ups

• arm64/convolve_neon.{c,h} were touched (rename only) but excluded from clang-tidy on x86 hosts because clang-tidy can't parse <arm_neon.h> intrinsics there. CI's lint workflow already excludes ^libvmaf/src/feature/arm64/.
• ADR-0141 §Historical debt's mention of _iqa_* reserved-identifier suppressions can be deleted once this lands (they don't exist).

🤖 Generated with Claude Code