Readability improvements to README

[kylebarron]

Changes

• Refer to the "proj" convention by its proper name geo-proj, not proj:. proj: is the key prefix and geo-proj seems to be the proper name of the convention
• More hyperlinks. There should be easy links to the geo-proj and multiscales conventions when we reference them.
• Move field attributes from h4 to h3 to make their definitions more prominent
• Change several "bolded headers" to "h4 headers". This is important because it gives an explicit anchor to these headers. So a consumer can link directly to "coordinate convention", for example
• Move the section about "implications of using spatial:transform with multiscales" as part of the definition of spatial:transform, not the example section.

[emmanuelmathot]

A few inline suggestions, plus two heads-ups:

• Stale proj: convention-name references remain outside the diff hunks: line 105 (bbox description), line 348 example heading (### Composition with proj: Convention) and intro, line 452 (Note how CRS information (proj:) is separated...), and FAQ entries at lines 491, 499, 501, 503, 507, 510 -- many of these use proj: to mean the convention, not the key prefix. Worth a follow-up pass to be consistent with this PR's stated rule.
• Likely conflicts with two open PRs touching the same sections: #11 rewrites the spatial:transform coordinate-convention / registration prose, and #24 restricts the schema and field descriptions to 2D and removes the 3D bullets in spatial:dimensions/spatial:bbox/spatial:shape. Probably easiest to land #11 and #24 first, then rebase this.

[maxrjones]

@emmanuelmathot @kylebarron would either of you have time to resolve Emmanuel's comments and sync this with the latest changes on main?

[kylebarron]

I think I've resolved all concerns and this should be good to merge