Revert "Add params ReadOnlySpan<T> overloads"

[jozkee]

Reverts #100898 in order to unblock dotnet/aspnetcore#55010 and dotnet/sdk#40189.

cc @AlekseyTs @MichalStrehovsky @akoeplinger @eerhardt @jeffhandley

[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, 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.

[dotnet-policy-service]

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

[akoeplinger]

/ba-g the failure in Build Analysis is a false-positive, the console log shows the test passed but there was some infrastructure error when uploading results to AzDO.

[stephentoub]

@jozkee, all the known issues were due to params, right? Could we resubmit this to add all the tests and all the new overloads, just without params on them? Once the language issues are addressed, we could then subsequently follow up to just add the keyword.

[jozkee]

As per dotnet/aspnetcore#55010 (comment), string.Join(string?, params ReadOnlySpan<string?>) would cause a source breaking change. What's our policy on that kind of changes?

cc @bartonjs.

[bartonjs]

From a guidelines perspective, StringValues shouldn't have had the implicit conversion operators, and then there wouldn't have been a problem :).

My guess is that we'll decide that this compile break is the right thing for the platform and resubmit; and anyone passing a StringValues as the only argument to Join or Format or what have you will just have to write in a cast to show which thing they meant... but that sounds like something we should decorate for API Review, mark it blocking, and discuss on Tuesday.

[stephentoub]

I agree with @bartonjs. However, even if it were an issue, that particular case is still only an issue because of the params, right? So for now we can still put back the new overloads, just without params, as I'd suggested?

[jozkee]

@stephentoub right, I just sent #101308 for that.