Initial port of RandomAccessStream/ReaderInfo

[kwhopper]

Port of drewnoakes/metadata-extractor-dotnet#130

The description and purpose are the same as in the C# pull request and issue:
drewnoakes/metadata-extractor-dotnet#131

Enhances or resolves the following issues:
#357
#351
#232 (support for this one, but commented out after the first SoS)

Some notable changes:

• The file or stream length is always required, either as an input or extracted from a seekable stream. This solves all kinds of EOF and semi-infinite loop problems.
• JpegSegmentReader implements the C# experimental approach to reading jpeg segments beyond the first SoS marker. This could be configurable in the future to allow all segments (including image data) to be accessible in some fashion.
• Just like the C# version, this PR corrects and records the global file/stream position of all tag data. This should allow future support for writing large byte arrays directly to disk or ignoring them altogether safely.

My Java skills aren't as refined as my C# skills. It would be great if someone could review. Thanks

[haumacher]

This would also allow fixing #149 and answering question #276 satisfactory?