Changing from negative-id CommentAnnotations and MapAnnotations to XMLAnnotations for non-OME data model info

[erickmartins]

This PR deals with issue #66 - we encode non-OME data model information (such as server path for files and provenance metadata) in XMLAnnotations instead of ad-hoc CommentAnnotations and MapAnnotations.

This allows us to provide and enforce schemas for these info. Right now, no schema validation is being done, but that's in the plans.

(PR work is still in progress - tests are not passing)

[joshmoore]

This makes me wonder what about existing workflows (such as those implemented during the recent hackathon) will break...

[erickmartins]

This makes me wonder what about existing workflows (such as those implemented during the recent hackathon) will break...

Anything that only uses the CLI interface should work as-is. But this will be a breaking change; packs generated with older versions will not be able to be unpacked with the new version.

[joshmoore]

nods At the hackathon folks were certainly starting to look into the XML. Knowing that this is why bioformats2raw is called such ("it's a black box!"), this might be something to discuss in the new year.

[erickmartins]

(as an FYI to folks with open PRs (@glyg @joshmoore @jburel ): I will merge this one last, because it's a big enough change that will mean releasing 1.0 after it!)