Add Unicode Blocks access and iteration

[sffc]

We should consider adding Unicode Blocks to the icu_properties crate. It should probably include:

1. Ability to loop over all blocks
2. Ability to access the code points in a block
3. Ability to look up a block by name

Concretely I think the most practical way to implement this would be to make an open enum for the block and then basically treat the block like an enumerated property, including code point access and display name parsing. I think it should pack pretty small in an InversionMap since the code point space is already segmented into fairly large contiguous blocks.

I imagine that the most difficult part of the implementation would be the sourcing of the data in datagen.

This would be a good first issue of medium scope.

Discussed in #4798

^{Originally posted by faassen April 11, 2024}
I dug around the source code, but I couldn't find a representation in Rust code of the Blocks.txt data. There's the unicode_blocks crate, but this misses an important feature; I need to be able to iterate through all blocks. I need to be able to look them up by name, but not the name as given as this has space characters. So iteration seems required so I can do some pre-processing. This is to implement regular expressions as defined by appendix F in the XML Schema specification:

Being able to get a CodePointInvList for a block would also be nice.

https://www.w3.org/TR/xmlschema-2/#regexs

Did I miss something? Is this planned?

[Manishearth]

One note is that we should document that unassigned code points may still report being in a block: which is not necessarily the behavior of all APIs I have seen that do this; but it is reasonable behavior. (And it's easy to get the other behavior by mixing this with Assigned/gc)