OpenSystemsDevelopment / qpdf

Also, delay adjustments for compressed objects until needed by
linearization checks.

Optimistically read subsection headers without reading individual object
entries, assuming that they are 20 bytes long as per the PDF spec. If
problems are encountered, fall back to calling bad_subsections.

Replace QPDFXRefEntry with a std::variant of structs.

Temporarily disable 3 specific-bugs tests. Remove 'xref size mismatch'
test.

Rename to Xref_table::process_section.
Process trailer processing subsection entries.

Calculate all subsections before reading subsection entries.

Duplicates some warnings for the time being.

Rename to Xref_table::subsection.
Return results instead of using out parameters.
Take on responsibility for throwing exception and calculation of offset of
first subsection entry.

Also, remove redundant clearing of the xref tabl during QPDF destruction.

To be encapsulated later.

Also, when recovering trailer from xref streams, pick the last valid
trailer encountered rather than the first.

Change the return type to a std::pair<bool, QPDFObjectHandle> in order to
allow a default constructed object handle (which is currently returned to
indicate failure) to become a valid object.

Tidy QPDF_Stream

Replace some std::shared_ptr parameters with reference parameters

1. Make class final
2. Pass og parameter by value
3. Properly initialize qpdf and og

Also, tweak QPDF::replaceObject to allow stream replacement without
violating the requirement that streams must always be indirect objects.

Also, remove QPDF::reserveStream as it does not do what the name implies
and having this as a separate methods does not aid code readability.

Given that the PDF spec requires the xref table to contain entries for all
object ids <= the maximum id present in a PDF document, max_size is a
qpdf implementation limitation for legitimate object ids.