Improvements to the text of RFC-3

[jni]

This PR updates RFC-3 to account for comments, questions, and data types raised
since it was created:

• There was discussion that RFC-3 is primarily for exotic use cases.
  I'll admit that I have some of those in mind 😂 but it is actually mostly
  going to serve standard microscopy images, as demonstrated by Zeiss's
  interest, as well as comment 3 from Cornelia Wetzker detailing their
  use of FLIM. In private communications, Cornelia mentioned that most of their
  data is in LIF (Leica) format, which supports 6D. I've tried to amend the
  text to front load these standard microscopy use cases.
• I've added FLIM to the data types that RFC-3 targets, to address Comment-3.
• I've added some user stories around some of the data types.
• I've added specifics to the proposal about how the spec's language would
  change.
• Finally, I've made two sample RFC-3 datasets using real-world data in
  Add two real datasets to the repo clbarnes/ome-zarr-rfc3-data#1 that can be used to test compliance with RFC-3.

Hopefully, this is enough to get this over the finish line soon...!

[github-actions]

Automated Review URLs

• Readthedocs

[joshmoore]

Thanks @jni 🙏 Would you mind also updating the version history at the bottom before merge?

[jni]

@joshmoore yup, did the thing!

[jo-mueller]

Looking cool! Just one more thing that came up during the discussion around ome/ngff-spec#130 (and this is only relevant because RFC3 happened to go in after RFC5):

I think there could (should?) be a statement about the discrete flag in the axes metadata, by which a dimension can declare whether interpolation is well-defined along itself. By extension, this can imform transformations how to treat all non-spatial axes.

[jni]

@jo-mueller what kind of statement? RFC-5 states axes/dimensions can be discrete or not, and RFC-3 has nothing more to say on the matter — we agree! 😂

[jni]

Having looked at it in more detail, the rc0 version of the spec states:

[axes metadata] MAY contain the field discrete. The value MUST be a boolean, and is true if the axis represents a discrete dimension (see below for details).

and

Arrays are inherently discrete (see Array coordinate systems, below) but are often used to store discrete samples of a continuous variable. The continuous values “in between” discrete samples can be retrieved using an interpolation method. If an axis is continuous ("discrete" : false), it indicates that interpolation is well-defined. Axes representing space and time are usually continuous. Similarly, joint interpolation across axes is well-defined only for axes of the same type. In contrast, discrete axes ("discrete" : true) may be indexed only by integers. Axes representing a channel, coordinate, or displacement are usually discrete.

I can reiterate that in the multiscale metadata section, but it would be redundant?

[jni]

@joshmoore as far as I'm concerned this is ready to be merged and sent out for (hopefully final) review! 🙏

[lubianat]

looks good!