Initial outline for allowing processing of in-memory buffers

[anjackson]

Hi @bitsgalore

This is an outline of a modification we would find useful for the BL Blitter work (see britishlibrary/blitter#4).

We'd like to be able to operate on in-memory data, but the current checkOneFile API is hard-coded to work on files. I've tried two different ways of implementing this, and this pull request seems to be the cleanest. But note I've not actually tested it yet!

Before testing it properly, I wanted to find out if you were happy to support in-memory analysis and if so, what you thought would be the cleanest way of implementing it?

(The other tactic I tried -- and used in production -- is here: https://github.com/britishlibrary/blitter/blob/master/jpylyzer/jpylyzer.py#L290-L318 . The problem with that approach is it makes handling errors more difficult).

[bitsgalore]

Hi @anjackson,

This looks like a useful improvement. Just let me know when you've completed testing, would be happy to accept a pull request along these lines.

[bitsgalore]

Closing because #162 takes care of this.