Added attribute to set raw data as the source of the pdf

[moesjarraf]

We used to pass in the pdf src as a {{ interpolation }} string. However this makes it really difficult to open a pdf from binary data.

This PR instead forces us to to pass in $scope variables, so we can pass the binary data to the directive. Note that interpolation strings will no longer be supported by the directive, so this is a breaking change.

Aside from passing in a URL string as the source, PDF.JS by default supports passing a Uint8Array as the source. Now we can actually use that feature in the directive.

It is fairly easy to create a UInt8Array from a pdf resource as can be seen in the updated demo file. Because you can just set the responseType of a XMLHttpRequest to arraybuffer and then pass that in the Uint8Array constructor.

I also fixed initializing the webviewer with no source. So you can actually load the pdf-viewer without displaying a pdf (for async loaded pdf files). It used to break the view if you passed in null as the src.

fixes #35 fixes #24

[jasny]

@moesjarraf Ok, but breaks BC.

Is there a way to do this without breaking BC? Like adding a new attribute data, while keeping src as is.

[moesjarraf]

@jasny I added a seperate attribute to not break BC. Also cleaned up instructions on how to use the directive in the readme. Removed unnecessary parts of the examples that just make the usage seem more complex.