Fix --normalize writing to wrong path with multiple files - #702

Summary

Fix incorrect file index when writing normalized output for multiple files.

Problem

When normalizing multiple files with normalizer --normalize file1.txt file2.txt, the CLI always references x_[0] (the first result entry) instead of the current file's result:

x_[0].unicode_path = join(dir_path, ".".join(o_))

with open(x_[0].unicode_path, "wb") as fp:
    fp.write(best_guess.output())

For the second file onward, this:

1. Overwrites the first entry's unicode_path with the current file's output path
2. Writes the normalized content to a path derived from the wrong entry

Fix

Use x_[-1] to reference the most recently appended result, which is always the current file being processed.