Add a new keyword to configure_file() that allows the developer to specify the input and output file encoding.
If the file encoding of the input is not UTF-8 meson can crash (see #1542). A crash as with UTF-16 is the best case and the worst meson will silently corrupt the output file for example with ISO-2022-JP. For additional details see pull request #3135.
The new keyword defaults to UTF-8 and the documentation strongly suggest to convert the file to UTF-8 when possible.