* Take decode arrays into account when processing grayscale images
* Adapt image extraction test case hashes
* Minor refactoring in the ColorAt image method
* Always return vanilla data from the jbig2 decoder
* Add ColorAt method for images
* Avoid resample on image to Go image conversion
* Avoid resample when converting grayscale image to RGB
* Preserve old behavior of image to Go image conversion
* Add missing case in the ToGoImage method
* Fix grayscale to RGB image conversion
* Improve code documentation
* Fix color extraction for CMYK and 4 bit RGB
* Add test case for the ColorAt image method
* Avoid resampling when converting CMYK image to RGB
* Add notice comment for the GetSamples/SetSamples image methods
* Add extract images test case, with memory profiling
* Use TotalAlloc insted of Alloc for memory profiling
* Remove calls to debug.FreeOSMemory from test cases
* Added text bounding box extraction.
* Add `font` field to textMark struct;
Create a new method `TextComponents` to retrieve all the text components of the extracted text in the page, with position and character informations
* Reorganizing extractor/text.go
* Added a text extraction position test.
* Added another text extraction location test.
* Text extraction location testing.
* Added tests for text extraction with location information.
* Cleaned up text extraction tests. No changes to functionality.
* Simplifying text extraction code.
* Simplified line construction in text.go
* Returning TextMark's in TextMarkArray which are based on PdfObjectArray but read-only, so not pointers.
* Added text extraction to show PDFs marked-up with bounding boxes of substring in extracted text.
* Add comments explaining how to calculate text bounding boxes.
* Made text_test.go naming consistent with function comments in text.go
* Use tm, pt, tl for textMark/TextMark PageText and TextLine receivers and local variables.
* uncommeted text stress test. Use go test --short to skip
* TextMark.Offset is now an index into the extracted text. It was an index into []rune(text)
Manually verified that output PDFs look good and leave hash check to detect change. If there is a change in the future, the hash change will trigger a failure upon which the output PDFs need to be re-checked and hashes updated if appropriate.
* Make PdfObjectDictionary Merge method chainable
* Update page resources Font dictionary when applying license information
* Add license font to the page resources only when it does not exist
* Update hash for split test after verification
