`Image()` signature/doc improvements

[fperez]

The Image call has proved to be remarkably popular, but its signature is a bit awkward.

• The most common usage I've seen from looking at many notebooks is to call it with a filename, but that's the third positional argument, forcing the vast majority of usage to be done with Image(filename=...), instead of simply Image(...).
• The data argument accepts raw data, a url or a filename, but the signature only mentions urls and data.

I think we should change the signature and dosctring to take simply a url/filename as the first argument, which could be called pathspec or location. For backwards compatibility we can continue accepting the keywords filename and url, but both aren't necessary.

The much less common usage of calling it with raw data can be relegated to be called with data = ..., though it's worth considering for backwards compatibility auto-detection of raw data if the argument is say longer than 10kbytes or so.

[minrk]

The data first positional arg is there because this model is uniform across all display classes:

DisplayClass(rawdata)

Now, it's easy to guess in the Image case if people are passing other info in the first position, and that's exactly what we do now (the docstring just hasn't kept up), but I do not think we should change the name or meaning of the first positional arg.

[minrk]

#3310 updates the docstring to be clearer. I can change the signature, too, if we really do want to go that way.

[minrk]

should have been autoclosed by #3310