For every issue string, json['issues'][issue_string] contains a list
of (page, count_of_issue_on_this_page) tuples.
To get the total number the issue appears in the current document,
we need to add up all the count_of_issue_on_this_page, not multiply the
page number with count_of_issue_on_this_page and add those (－‸ლ)
* Count files that are password-protected
* Use `pdf --json` to groups by missing feature and then by file
instead of by file first, feature second
* Count files that render with issues
* Print number of files without issues last
* Always print all crash stacks
* Elide parser offsets to better group parser errors
* Use `backslashreplace` for decoding crash stacks so that we don't
crash when printing crash stacks if the error output isn't valid
utf-8
* Swap last two lines of output, reads a bit better
The rendering happens only in-memory, so this is only useful for
looking at the crash rate and the reports of missing features.
To actually see the output of a file, use
pdf --render out.png --page N path/to/input.pdf
instead.
