PLAYA-PDF 0.6.0: Structure and text improvements

What's Changed

• Iterate over Form XObjects inside Form XObjects with .xobjects by @dhdaines in #126
• Correct bbox on non-diagonal Type3 FontMatrix by @dhdaines in #127
• Fixes and improvements to text extraction, marked content and logical structure by @dhdaines in #124
• Add displacement property to text objects by @dhdaines in #129
• Allow iteration over Type3 font programs by @dhdaines in #130
• Extract images as PNMs if possible by @dhdaines in #131

Notes from CHANGELOG.md

• Add structure to Page to access structure elements indexed by
  marked content IDs (convenience wrapper over the parent tree)
• Add structure to XObjectObject for the same reason
• Add parent to all ContentObject to access parent structure
  element (if any) via the parent tree
• Descend into Form XObjects in Page.xobjects
• Improve text extraction for rotated pages
• Improve text extraction for tagged PDFs
• Correct displacement and bbox for Type3 fonts with non-diagonal
  FontMatrix
• Add displacement property to TextObject
• Add functioning __iter__ to GlyphObject in the case of
  Type3 fonts, which works like XObjectObject
• Extract non-JPEG images as PNM
• BREAKING: Fix __len__ on PathObject which incorrectly returned
  non-zero even though iteration is not possible
• BREAKING: Remove misleading char_width, get_descent, and
  get_ascent methods and hscale and vscale properties from font
  objects
• BREAKING: Do not guess basename for Type3 fonts (generally it
  isn't different from fontname for other subset fonts)
• BREAKING: Element.contents contains both structure.ContentItem
  and structure.ContentObject

Full Changelog: v0.5.1...v0.6.0