src: use simdutf for UTF-8 ⇄ UTF-16 transcoding in StringDecoder and StringBytes::Write - #65324

StringDecoder gets 3–12× faster on UTF-8, and writing non-ASCII strings as UTF-8 (Buffer#write, Buffer.from(string),
every stream/socket/fs string write) gets 4–5× faster. A 64 KiB JSON-RPC round trip over a child's stdio (stringify → write
→ readline → parse, non-ASCII payload) drops from 1.28 ms to 0.90 ms with only the parent patched, 0.67 ms with both ends.

string_decoder/string-decoder.js encoding='utf8' inLen=1024 chunkLen=1024      ***  1125.82 %  ±8.25%
string_decoder/string-decoder.js encoding='utf8' inLen=128  chunkLen=1024      ***   322.79 %  ±2.28%
string_decoder/string-decoder.js encoding='utf8' inLen=32   chunkLen=1024      ***    98.53 %  ±1.37%
string_decoder/string-decoder.js encoding='utf8' inLen=1024 chunkLen=16        ***     5.15 %  ±0.58%
buffers/buffer-write-string-utf8.js (new)  len=65536 chars='two-byte'                 ***   391.17 %
buffers/buffer-write-string-utf8.js        len=65536 chars='two-byte-lone-surrogate'  ***   314.76 %
buffers/buffer-write-string-utf8.js        len=256   chars='two-byte'                 ***   289.28 %
buffers/buffer-write-string-utf8.js        len=256   chars='two-byte-astral'          ***   228.97 %
one-byte strings / other encodings                                                             ~0 %  n.s.

(Linux x64, benchmark/compare.js, 30 runs, significance as in compare.R.)

Two commits:

string_decoder: decode UTF-8 via StringBytes::Encode - the decoder built strings with String::NewFromUtf8();
buffer.toString() already goes through StringBytes::Encode(), which uses simdutf. MakeString() now calls the same
function (the kMaxLength → ERR_STRING_TOO_LONG check stays in front), so the decoder returns exactly what toString()
returns for the same bytes. The partial-character bookkeeping is untouched.

src: use simdutf for two-byte strings in UTF-8 writes - StringBytes::Write(UTF8) used String::WriteUtf8V2() for
two-byte strings. For strings longer than 32 code units it now validates with simdutf::validate_utf16() (lone surrogates
are repaired into a scratch buffer with to_well_formed_utf16(), so the output stays byte-identical to V8's
kReplaceInvalidUtf8) and transcodes with convert_utf16_to_utf8() - but only when the destination is known to fit
(buflen >= 3 × length or >= utf8_length_from_utf16()); otherwise it falls back to V8, so partial writes truncate at the
same character boundary as before. Shorter strings and one-byte strings keep the V8 path.

Tests: test-string-decoder-utf8-large.js (new: large inputs, chunk boundaries inside multi-byte sequences, invalid
sequences vs toString()); test-buffer-write-utf8-two-byte.js (new: independent reference encoder; BMP/astral/lone
surrogates at start/middle/end; exact-fit, one-short and 3× destinations; partial writes). Existing string_decoder, buffer,
stream, net, http, child_process and readline suites pass.

Disclosure: the code, test, measurements and this description were written by Claude Code, directed and reviewed by @codebytere.