Convert Java modified UTF-8 to UTF-16 following ICU 78.1 ustrtrns.cpp.
One to three byte sequences are decoded without the overlong and
surrogate checks of standard UTF-8, so C0 80 is U+0000 and encoded
surrogates pass through; four byte sequences and truncated or malformed
ones are replaced by the substitution character, or fail with
U_INVALID_CHAR_FOUND when it is U_SENTINEL. The bytes a malformed sequence
swallows follow ICU's utf8_nextCharSafeBody. As in ICU, once the
destination is full each substituted sequence adds one unit to the
preflight length even when the substitution character is supplementary.
The sceIcuUnknown stub with this NID is removed.
