Fixes #258, #259

• Added support for PDF text rendering mode (the Tr operator). This makes it possible to insert invisible text or glyphs with bordered and filled lines.
• Optimize support for JPX image extraction from PDFs. For these images, we now directly read the binary data.

Fixes JBIG2 decode error

• JBIG2 decode error impacted Document.extractImage and Page.getText method in rare scenarios
• new Pixmap.setRect method
• extend several Pixmap methods to also support non-alpha images

Minor feature improvements and bug fixes

• new Pixmap.setPixel() method
• documentation: new How-To on image conversion
• new text length calculation for Base-14 fonts
• new Postscript output for pixmaps
• Pixmap.writeImage() by default determines image type from filename extension

Extending Draw Methods and Pixmap Output Options

Merge pull request #245 from JorjMcKie/master

Upgrade to v1.14.5

Fix #239

Merge pull request #240 from JorjMcKie/master

merge v1.14.4

Supporting CJK Font Text Insertion

Merge pull request #233 from JorjMcKie/master

merge v1.14.3

Fixes #223 - Pathlib Support for Document class

Fixes #227
Also implemented several small performance improvements.

Supporting MuPDF v1.14.0

This version comes with major changes on the technical level and hardly visible to the user. Among those low-level changes are internal handling of geometry object (rectangle, points and friends) and name changes to standard PDF objects, like PDF_NAME_Rect has been renamed to PDF_NAME(Rect).

But there are notable functionality enhancements as well:

• updating an encrypted PDF lets you optionally keep encryption
• MuPDF messages going to STDOUT and STDERR are now intercepted and will no longer appear
• several new annotation types are officially supported by MuPDF
• a new geometry object Quad supports rectangles that are no longer axis-aligned. This allows addressing location of "tilted" text (i.e. which is neither horizontal nor vertical)

As always, a number of issues has been fixed as well.

Improved E-Book Support

You can now re-layout documents and trace changed page numbers.

Introducing Python Memory Management

For Python version 3 builds, we are introducing Python memory management. This means that all memory (de-) allocation calls are passed to the Python memory manager - whether they are issued by PyMuPDF's inface C code or issued by the base library MuDPF.
This should lead to an improved memory usage profile, because Python has complete control over all memory used. On Windows, we have also seen performance improvements (about 2% and more) for important basic use cases like text extraction and pixmap creation.

We have also worked on the new "Recipes" chapter of the documentation and removed 2 bugs occurring on Python 2 platforms only.