ANN: pandoc 2.1

[John MacFarlane <jgm-TVLZxgkOlNX2fBVCVOL8/A@public.gmane.org>]

I'm pleased to announce the release of pandoc 2.1.

Binaries & changelog: https://github.com/jgm/pandoc/releases/tag/2.1
Source & API docs: https://hackage.haskell.org/package/pandoc

This release increments the major version number because
of a few incompatible API changes; however, they are fairly
minor and should not affect most users of the library.

* In Text.Pandoc.Class, the FileTree constructor is no longer exported;
  use insertInFileTree and getFileInfo to modify a FileTree.
* In Text.Pandoc.App, the type of applyFilters has changed, and
  applyLuaFilters has been removed. The Filter type is also now exported.

This last change gives us a user-facing improvement:  JSON
filters and lua filters can now be run in any order.  (They
will be run in the order they appear on the command line.)
Previously all lua filters ran before all JSON filters.

A major performance regression in earlier 2.x versions has
been fixed.  This caused the pandoc executable to run
much slower at startup.

Albert Krewinkel has continued to improve the lua filters
system, with many improvements that make it easier to write
lua filters.

Jesse Rosenthal has improved the Powerpoint writer and
fixed some issues with the docx reader.

Alexander Krotov has continued to improve the Muse reader
and writer.

Sebastian Talmon has fixed issues with subtitles and
headerless tables in the docx writer.

oltolm has improved formatting of formulas in
the OpenDocument/ODT writer.

Thanks to everyone who contributed with code, discussion, or
bug reports.