[BUG] - Stream: false table detection

[aropb]

When switching to PdfPig 0.1.15, a false table detection appeared on the page where it does not exist.

[BobLd]

@aropb do you have a sample document?

[aropb]

I have attached the document:
Indonesia Overhauls Its Financial Sector_ An Introduction to the Amended P2SK Law _ Rajah & Tann Asia.pdf

Yes, it looks strange, the text goes beyond the boundaries of the page. But at least the old version doesn't create a huge number of empty tables.

I also came to the conclusion that I have to use both methods of extracting tables at once (stream, spreadshet). Then I combine them, excluding duplicates (priority for stream). Because in some cases, one method works and the other does not. Of course it's expensive. I wish I had one method that tests the page for both detection and extraction methods at once.

            var streamTables = GetFilteredTables(page.GetTablesStream());
            var latticeTables = GetFilteredTables(page.GetTablesLattice());

            return streamTables.Union(latticeTables.Where(x => !streamTables.Any(y => x.Intersects(y))))
                .OrderByDescending(x => x.BoundingBox.Top)
                .Select(x => new TableDesc
                {
                    Table = x
                })
                .ToList();