WIP: Extend API with "presentation state" support - DO NOT MERGE

[forderud]

This will expose a sorted list of slice/rendering "views" that clients can utilize to reconstruct slice/rendering views that are similar to the views shown on screen when saving the original data.

Out of scope for now:

• The on-screen layout is currently not exposed. Clients will need to handle positioning themselves, which might not match the original layout.
• Perspective projections is not currently supported, only a orthographic cube. Clip-planes is similarly not supported either. Implementers will therefore need to expose the orthographic representation that most closely resembles what is displayed on the original system.

Views are encoded with the following parameters:

• Origin: x,y,z (top-left corner of the view)
• dir1: x,y,z (directional vector from top-left to top-right corner)
• dir2: x.y,z (directional vector from top-left to bottom-right corner)
• dir3: x.y,z (directional vector from front-top-left to back-top-left corner). Zero on slices.

Assumptions:

• It's possible to somehow infer the default slice/rendering "views" from a rawdicom file.
• It's possible to dynamically change the "views" in the PACS system based on this metadata.

Example

[forderud]

Have now published the proposal on MedicalUltrasound#141 . Therefore, closing this PR.