Fixed missing zero terminating for strings, fixed conversion using UnicodeToUtf8, unicode2ascii

• Fixed conversion of string constants when using unicode2ascii, Utf8ToUnicode, UnicodeToUtf8.
  The UnicodeToUtf8 function writes 0 to the end of the string, and memory must be allocated for this 0.
  The Utf8ToUnicode function does not write 0 to the end of the string, but, like UnicodeToUtf8, it returns +1 to the length of the resulting string.
  The unicode2ascii function does not write 0 to the end of the string, it returns the length of the string without the terminating 0.

• In tstringconstnode, set the len after converting ascii2unicode.
  len is used in tcgstringconstnode.pass_generate_code as the HashTable key length,
  used in ppuwrite for the Attribute constructor parameter value.

• Fixed a potential bug in the scanner when reading a string constant expression.
  The s: string argument is of type shortstring, i.e. s is not zero-terminated.

• In TElfObjInput.ReadObjData also needs to terminate the strings shstrtab, strtab with zero.
  Judging by the string(PChar(@shstrtab[shdr.sh_name])) code, the terminating zero is needed.

• Fixed conversion from UTF16 to UTF8 in the TStringConstNode.XMLPrintNodeData method.
  The number of UTF8 bytes can be greater than the length of the UTF16 string.

Also at the same time

• Fixed using UnicodeToUtf8 in fpwidestring unit (issue #39746)
• Fixed using UnicodeToUtf8 in android/cwstring.pp, nativent/sysutils.pp

Resolves #39746