refactor(core): C++23 pilot Wave 1 — opt.c → opt.cpp (ADR-0721)

[lusoris]

Summary

Converts core/src/opt.c → core/src/opt.cpp using C++23 std::optional<T> for the
three internal parse helpers (parse_bool, parse_int, parse_double). The public C
ABI (vmaf_option_set) is identical — same signature, same return contract, same
case-sensitive string comparison behavior. opt.h gains extern "C" guards. Follows
the ADR-0708 playbook for metadata_handler.cpp (C++20 Wave 0 pilot).

Also fixes a pre-existing meson setup failure: test_ansnr_simd.c was deleted in
ADR-0720/PR #38 but its core/test/meson.build stanza was not cleaned up.

Type

• refactor — no behavior change

Checklist

• Commits follow Conventional Commits (the commit-msg hook enforces this).
• make format && make lint is green locally.
• Unit tests pass: meson test -C build-cpp23-pilot --suite=fast — 50/50 PASS.
• If I touched any SIMD/GPU code path, I ran /cross-backend-diff and the worst ULP is ≤ 2.
  no SIMD/GPU code path touched — pure C→C++ internal refactor.
• If I touched a feature extractor with SIMD/GPU twins, I either updated every twin or listed the gap under "Known follow-ups" below.
  no feature extractor touched.
• If I added a new .c / .cpp / .cu / .h / .hpp, it has the appropriate license header.
  core/src/opt.cpp carries the Netflix BSD+Patent header (upstream-touched file).
• If this is a breaking change, the commit message uses ! or BREAKING CHANGE:.
  no breaking change.
• If this PR adds an ADR, the ADR row lives in docs/adr/_index_fragments/<NNNN-slug>.md and the slug is appended to docs/adr/_index_fragments/_order.txt.
  docs/adr/_index_fragments/0721-cpp23-pilot-opt.md + _order.txt updated; scripts/docs/concat-adr-index.sh --write run.

Bug-status hygiene (ADR-0165)

no state delta: pure internal refactor with no bug opened, closed, or ruled out.

Netflix golden-data gate (ADR-0024)

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

Cross-backend numerical results

no SIMD/GPU path touched; opt.c is pure string→scalar parsing with no arithmetic output.

Deep-dive deliverables (ADR-0108)

• Research digest — no digest needed: mechanical conversion following ADR-0708 playbook; no novel design decisions.
• Decision matrix — captured in ADR-0721 ## Alternatives considered (expected vs optional, string_view vs const char*, global cpp_std bump vs per-target override).
• AGENTS.md invariant note — added to core/AGENTS.md: opt.cpp upstream-sync contract and extern "C" guard preservation.
• Reproducer / smoke-test command — see Reproducer section below.
• CHANGELOG fragment — changelog.d/changed/cpp23-opt.md.
• Rebase note — docs/rebase-notes.md §core/src/opt.cpp — C++23 Wave 1 pilot (ADR-0721).

Reproducer

# Build host-side, CPU only
meson setup build-cpp23-pilot core -Denable_cuda=false -Denable_sycl=false
ninja -C build-cpp23-pilot

# Fast test suite — must be 50/50
meson test -C build-cpp23-pilot --suite=fast

# CLI smoke — flag list must be present
./build-cpp23-pilot/tools/vmaf 2>&1 | head -10
# Expected: Usage: ... options including --model, --feature, --reference, etc.

no user-discoverable surface change — internal refactor; CLI parse is bit-equivalent (verified via --help smoke).

Known follow-ups

• Wave 2: log.c (ROI 3.0) and mem.c (ROI 2.0) — separate PR.
• dict.c (Wave 3, ROI 0.83) deferred until cpp_std=c++23 is set project-wide (requires std::expected).
• std::expected<T, int> as a replacement for std::optional + separate -EINVAL return: deferred until the host toolchain ships a stable <expected> (gcc ≥ 13 / clang ≥ 16 required; current CI uses gcc 12/clang 15).

[lusoris]

CRITICAL finding from adversarial review: opt.cpp parse_int/parse_double call std::strtol/std::strtod via sv.data() without guaranteed NUL-termination. If any future caller constructs a string_view over a substring (not a full C-string), strtol reads past the slice boundary — UB. Fix: copy sv to std::string before calling strtol/strtod, or add assert(sv.data()[sv.size()] == '\0'). See docs/research/cpp23-wave-adversarial-review-20260528.md finding #3.

[lusoris]

Fix for CRITICAL adversarial review finding in PR #78 pushed as 59cbf81. See PR #78 digest for context.