fix: fix negative index error on RemoveUnusedNode when open pdf with …

[juniorschroder]

Hello @VahidN, how are you?
We are using your fork of iTextSharp in our software and have encountered a very specific issue. We have a PDF document that contains a link element with a reference that appears to be invalid:

This does not compromise the PDF itself, but it generates an error when opening the PDF with the PdfReader class:

I am opening this pull request with the correction made. Unfortunately, I couldn't find another PDF with the same problem, and the PDF I have is owned by a client, so I don't have permission to share it. For this reason, I am not including a test along with the code.

[what-the-diff]

PR Summary

• Improved Condition Checking in 'RemoveUnusedNode'
  The conditions used within this method in PdfReader.cs were refined. The reasoning behind this change is to ensure accuracy when performing actions depending on these conditions. Previously it just checked if a particular node was accessed (!hits[num]), now the condition also checks if num is greater than 0 (num > 0 && !hits[num]). This adds another layer of verification before proceeding with further actions.

• Enhanced 'num' Validations in 'RemoveUnusedNode' Method
  The modified conditions now incorporate additional validation related to num. The updated condition extends the existing check for the range and 'null' state of _xrefObj[num] to also ensure that num > 0. This adjustment not only prevents potential errors from handling invalid data but also correctly handles edge cases involving _xrefObj[num] where num equals 0.