31: Load data from URIs

[krzywon]

This PR adds a few non-breaking enhancements to the data loader system to aid in scripting and web-facing capabilities.

• Fixes load data from url #31: The Loader class now recognizes URI strings, and, using the python built-in module urllib.request, opens the file at from the web address.
• Refs Load data button should also be able to folders sasview#1845: Lists of files can now be sent to the loader instead of just a single file and all loaded Data objects will be returned as a list. Note: This functionality is not available to individual readers.
• The loader is directly callable to allow transient instances of the loader object to aid in scripting.
• File handles are passed to the readers instead of file names so data files (especially remote files) are only opened a single time, regardless of the number of loaders attempting to read the file, while maintaining backward compatibility when passing a file path directly to a reader.
• Unit tests were added
  • Comparing file contents loaded from the local repository to files loaded from their GitHub URIs for the following formats: ABS, Anton-Parr SAXSess PDH, canSAS XML, NXcanSAS, 2D NIST DAT, and SESANS.
  • Comparing file contents loaded individually as compared to files loaded as a list.
  • Ensuring Readers and Loader behavior matched for individual files
• The ReadMe was updated to match the new functionality.

[pkienzle]

Tested with pytest and python -m sasmodels.compare sphere -data=https://github.com/SasView/sasdata/raw/master/test/sasdataloader/data/valid_cansas_xml.xml

I still think the CustomFileOpen is unnecessarily complicated but it works as is, and it is just as much effort to change in a future PR as it is to change in this PR.