Add the approved span marshallers and remove the v1 model

[jkoritzinsky]

This PR builds on #71978 and adds the approved span marshaller shapes. Since we will no longer have usage of the v1 marshalling model, this PR also removes it entirely.

This PR would replace #71980, but that one should be merged if this one isn't ready by P7. If this PR misses P7, we need to add type forwarders for the types moved between the ref assemblies.

In combination with the referenced PRs, this PR fixes #70859

Fixes #69281

[ghost]

Tagging subscribers to this area: @dotnet/interop-contrib
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR builds on #71978 and adds the approved span marshaller shapes. Since we will no longer have usage of the v1 marshalling model, this PR also removes it entirely.

This PR would replace #71980, but that one should be merged if this one isn't ready by P7. If this PR misses P7, we need to add type forwarders for the types moved between the ref assemblies.

In combination with the referenced PRs, this PR fixes #70859

Fixes #69281

Author:	jkoritzinsky
Assignees:	-
Labels:	area-System.Runtime.InteropServices, blocked, source-generator
Milestone:	-

[dotnet-issue-labeler]

Note regarding the new-api-needs-documentation label:

This serves as a reminder for when your PR is modifying a ref *.cs file and adding/modifying public APIs, to please make sure the API implementation in the src *.cs file is documented with triple slash comments, so the PR reviewers can sign off that change.

[stephentoub]

@jkoritzinsky, so once this PR is in, we can start using spans (unadorned) in LibraryImport methods?

[jkoritzinsky]

@stephentoub Yes! I'd update places to use this now, but I want to try to get this in for P7 if possible.

[jkoritzinsky]

The failing test should be deleted. I have a commit locally. That deletes the test, but I'm going to let CI keep running to see if there's anything else that fails. If this is the only failure, I'll push up the change to delete the test and then merge the PR.

That way we don't get this PR stuck waiting for machines to run tests when we want to get it in before P7 branches off.

[jkoritzinsky]

Okay all of the test failures are the expected failure.