[release/9.0-staging] Fix analyzer tracking of nullable enums

[github-actions]

Backport of #109430 to release/9.0-staging

/cc @sbomer

Customer Impact

• Customer reported
• Found internally

Originally fixed in .NET 10 when found while investigating a related issue, and we got a customer report after the .NET 9 release. Customer reported via email when attempting to make an app NativeAOT and trim compatible. Impact is that there are unexpected analyzer warnings in source-generated code.

Simple example of a false positive warning customers could hit:

usingSystem.Diagnostics.CodeAnalysis;usingSystem.Runtime.CompilerServices;RuntimeHelpers.RunClassConstructor(typeof(Nullable<E>).TypeHandle);// unexpected warningenumE{}

The actual customer-reported issue showed up in source-generated code not controlled by the customer, in a WinUI app:

IL2059: Unrecognized value passed to the parameter 'type' of method 'System.Runtime.CompilerServices.RuntimeHelpers.RunClassConstructor(RuntimeTypeHandle)'. It's not possible to guarantee the availability of the target static constructor.

for generated code like:

privatevoidStaticInitializer_158_Nullable()=>global::System.Runtime.CompilerServices.RuntimeHelpers.RunClassConstructor(typeof(global::System.Nullable<global::CommunityToolkit.WinUI.Animations.EasingType>).TypeHandle);

Regression

• No

Testing

Added unit test which validates that the same pattern reported by the customer does not produces a warning.

Risk

Low. Analyzer-only change, doesn't impact runtime. Removes incorrect warnings produced by the trim analyzer.

[dotnet-policy-service]

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

[sbomer]

This is surfacing existing test failures, fixing in #110340.

[jeffschwMSFT]

lgtm. please get a code review. we will take for consideration in 9.0.x