Move the check for encryption into PdfParser instead of the CrossReference class

[JanSlabon]

To place the logic in the CrossReference class looks somewhat odd. We should refactor this.

[JanSlabon]

And throw a more context specific exception: e.g. UnsupportedException.

[JanSlabon]

The move is actually impossible because we need to check for encryption when we reach a compressed cross reference. If we would not check for this a missleading exception will be thrown, that would say that the file can be processed with the FPDI PDF-Parser add-on, which is not true.

The new exception type is postponed to v3.