Find the stop byte of a string run without a byte loop

[nlohmann]

Part of the stack for the zero-copy view (#5295). This change speeds up json::parse and dump() on its own; the view uses the same kernels later.

Summary

The string kernels in detail/input/string_scan.hpp test eight bytes at a time. Until now, once a word contained a byte to stop at, the kernels found that byte with a byte loop. The lowest flagged byte of the SWAR tests is always a real hit, because the borrows of the subtractions can only flag bytes above one. So its index is now the trailing-zero count of the mask.

Words are assembled in little-endian order on every platform, so the result doesn't depend on the byte order. count_trailing_zeros() has a portable fallback.

Non-ASCII text: scalar_string_bulk_run() now validates a run of multi-byte UTF-8 sequences one after another, instead of searching for the next special byte after each sequence. This helps text in non-Latin scripts.

The kernels serve the lexer's contiguous fast path, the serializer, and the binary formats.

Tests

• New tests compare all three kernels with byte-by-byte reference scans on 100,000 generated buffers, at three alignments.
• The portable fallback of count_trailing_zeros() is checked against the builtin.
• The existing suite passes under AddressSanitizer and UndefinedBehaviorSanitizer.

Benchmarks

tests/benchmarks: Apple M1 Max, Apple clang 21, -O3 -flto, median of 2×5 interleaved repetitions:

benchmark	change
ParseString/twitter	−7.1%
ParseString/jeopardy	−4.2%
ParseIndented/twitter	−8.3%
Dump/twitter	−16%
Dump/jeopardy	−7.2%
Dump/citm_catalog	−4.7%
other ParseString/Dump rows	within ±2%

ParseFile reads from a stream and does not use these kernels; its rows are unchanged (within noise).

json::parse in a separate process, best of 5:

file	change
poet.json (Chinese text)	−27%
random.json	−7.6%
gsoc-2018.json	−6.2%
twitter.json	−5.6%
each of the other 10 corpus files	0.4–4.6% faster

Public API

No change.

---

Written by Claude Code.

🤖 Generated with Claude Code

[nlohmann]

CI fixes for this PR (commit 046ad7e, plus the fixes from #5617 merged in):

• ci_test_gcc, ci_test_standards_gcc (11): -Werror=useless-cast on static_cast<std::size_t>(next() % n) in the string-scan test of unit-class_lexer.cpp. On 64-bit Linux, std::uint64_t and std::size_t are the same type, while the cast is needed where std::size_t is 32 bits. The sizes now come from a 32-bit value (next_small()), which converts to std::size_t implicitly on every platform.
• ci_clang_tidy: inherited from Convert floats with Eisel-Lemire when std::from_chars is unavailable #5617 and fixed there.

Verified with GCC 16 using the CI flags (the test passes).

#5619 (json-view/04-unicode-escapes) got only a propagation merge; its failures were all inherited from here and from #5617.

— posted by Claude Code on behalf of @nlohmann

[nlohmann]

The zero-copy view stack (#5295) was rebased onto develop and regrouped from 20 PRs into 10, one squashed commit each. The changes of this PR are now in #5738; its review comments stay here for reference.

---

Written by Claude Code.