Currently we detect a graphic format in a ImpPeekGraphicFormat
method, which is called when reading a graphic from a stream.
The code is quite convoluted and doesn't help with readability
of graphicfilter.cxx

Additionally there exists an detection code written for the UNO
class GraphicDescriptor, which duplicates the detection, but can
in addition also extract other metadata at the same time.

This introduces the initial implementation of GraphicFormatDetector
class. It will first replace the code in ImpPeekGraphicFormat
to detect the graphic format and then later also be extended to
do what the GraphicDescriptor is doing. But currently it only
duplicates the implementation of ImpPeekGraphicFormat.

The problem with both of the current solution is that there are
not any tests written. For the GraphicFormatDetector however the
approach is to add test cases to check the current implementation
and only then refactor the code.

Change-Id: Idb9d3859b4380e3f15237d97aff969ce81e631dd
Reviewed-on: https://gerrit.libreoffice.org/72086
Tested-by: Jenkins
Reviewed-by: Tomaž Vajngerl <quikee@gmail.com>