Fix delimiting within <choose> and <layout> for CSL

[YDX-2147483647]

• Fix delimiting within <choose> and <layout> for CSL

  Fixes The delimiter attribute of <group> is not applied to items inside <choose> #180

  Fixes Issue with CSL Rendering (OSCOLA no ibid Style) #109

  • Before: Frank Mittelbach and others, The Latex Companion (2nd ednAddison-Wesley Professional 2004).
  • After: Frank Mittelbach and others, The Latex Companion (2nd edn, Addison-Wesley Professional 2004).
  • Diff: The comma after 2nd edn.

• Add a minimal test case

  • Before: Title_Edition_PublisherPlace_TitleEditionPublisherPlace
  • After: Title_Edition_Publisher_Place_TitleEditionPublisherPlace
  • Diff: _ between the first Publisher Place.

There is a design trade-off:
The delimiter stack uses String instead of &str.
Using &str could eliminate many cloning, but at the cost of introducing complex lifetimes.
Considering that delimiters are typically short, I choose String.

Besides, I bump some *.cbor to pass tests. If this is not desired, I can reset the commit.

[PgBiel]

Thank you!

[YDX-2147483647]

You are welcome!

And I notice that you updated the doc comments for push/pop (7a341cc).

(the delimiter) is inherited by [citationberg::Choose] for its children as well.

The world might be a bit confusing. In XML, the term_child_ only includes the direct ones. (The term for both direct and indirect ones is descendant.)
The spec uses output instead:

delimiting the output of the child elements

applied within the output of cs:choose

[PgBiel]

It is correct in the sense that it only passes the delimiter forward to children, which may decide to pass it further down the tree or not - usually not, as the point of delimiter is to delimit only the immediate children of a group or layout, unless the child of the choose is another choose. The usage of the word "output" seems to be there more to indicate that it won't apply to, say, the <else> branch if it wasn't matched, but otherwise it will apply to the final children.

So I guess it's more a matter of how you interpret it - it's true that the delimiter may affect descendants other than children with nested <choose>, but I also wasn't looking to summarize how the children handle delimiters.

Either way, we could adopt a wording closer to the spec to be sure.