Gracefully handle the is_aes check without throwing errors

[StephanvanSchaik]

Commit 9f22915 introduces a regression in 0.35 compared to 0.34 when decrypting a PDF document that does not use AES. Instead of decrypting the document, it throws an error that the CF dictionary is missing. Consequently, this prevents the document from being decrypted.

To remedy this, this pull request eliminates all uses of ? in the is_aes check to gracefully determine if the PDF document uses AES or not. It then proceeds to decrypt it accordingly.

Instead of using ?, we carefully convert all Result<T, E> types into Option<T> using .ok() in combination with .map() and .flatten() to walk through the series of dictionaries. Then at the final step we simply compare if the cfm is equal to b"AESV2". If any of the steps fails, we would get None, in which case we default to is_aes = false (rather than throwing an error).

[genusistimelord]

I approve these changes. @J-F-Liu I think this is acceptable and does indeed fix the encryption issue.

[J-F-Liu]

Please use and_then to fix clippy warnings

[StephanvanSchaik]

I just pushed a commit addressing the clippy lints.