[release/9.0] Fix Encoding regression

[github-actions]

Fixes #110521

Backport of #111320 to release/9.0

/cc @tarekgh

Customer Impact

• Customer reported
• Found internally

In .NET 9.0, users working with multi-byte encodings like GB18030 may encounter unexpected exceptions when decoding bytes into a Unicode surrogate pair. It's important to note that encoding can be used implicitly, such as when reading streams, which significantly broadens the potential impact of this issue.

Regression

• Yes
• No

This is a regression in .NET 9.0 from the change #97950. The change has modified the logic in EncodingCharBuffer.AddChar to handle uninitialized chars differently causing regression.

Testing

The regression tests have passed. I manually debugged the specific scenario and confirmed that the fix is effective. Additionally, I added a test case to cover the previous failure, ensuring that this regression will be detected if it occurs again in the future.

Risk

Low, we didn't change any logic more than handling the failing case when dealing with null buffers.

IMPORTANT: If this backport is for a servicing release, please verify that:

• The PR target branch is release/X.0-staging, not release/X.0.

Package authoring no longer needed in .NET 9

IMPORTANT: Starting with .NET 9, you no longer need to edit a NuGet package's csproj to enable building and bump the version.
Keep in mind that we still need package authoring in .NET 8 and older versions.

[dotnet-policy-service]

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

[tarekgh]

CC @ericstj @artl93

[tarekgh]

/ba-g the failure is a build timeout which is seen in other PRs too.