fix off-by-one null terminator in toUTF8_UTF16

[zayeem06]

Off-by-one terminator in toUTF8_UTF16
The UTF-16 to UTF-8 converter in runtime/C/src/antlr3string.c writes its terminator at *(outputEnd+1) after ConvertUTF16toUTF8 leaves outputEnd one past the last byte, so it lands a byte too far. On an empty UTF-16 string the buffer is a single byte (size = len*3 = 0, malloc(1)) and the write goes to index 1, one byte past the allocation; on non-empty input it leaves a stray uninitialized byte in front of the terminator. Terminating at *outputEnd writes at the byte the converter actually stopped on, which is always inside the size+1 allocation.

Reviewers may want to confirm the empty-string path: newStr8("")->toUTF8 reaches this with len == 0. ASan flags a 1-byte heap-buffer-overflow write there before the change and is clean after.

[zayeem06]

any update on this?