Delete more leftover UTF-16 code from compiler

[MichalStrehovsky]

No description provided.

[Copilot]

Pull request overview

This PR updates NativeAOT’s compiler name mangling to further move string-literal mangling onto Utf8String (removing more UTF-16-centric code paths), and adds a NativeAOT smoke test to validate correct handling of isolated UTF-16 surrogate literals.

Changes:

• Add a new NativeAOT smoke test (MiscTests) that validates string literals containing isolated surrogates are preserved as single char code units.
• Change NameMangler.GetMangledStringName(string) to return Utf8String and update call sites (e.g., NodeFactory.ConstantUtf8String) accordingly.
• Adjust NativeAotNameMangler string-literal hashing to be primarily UTF-8-based, with a special-case for literals whose UTF-8 encoding contains U+FFFD replacement bytes.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/tests/nativeaot/SmokeTests/UnitTests/UnitTests.csproj	Adds the new MiscTests.cs compile item to the NativeAOT smoke tests project.
src/tests/nativeaot/SmokeTests/UnitTests/MiscTests.cs	New test validating isolated surrogate string literal preservation.
src/tests/nativeaot/SmokeTests/UnitTests/Main.cs	Wires MiscTests.Run into the smoke test runner.
src/coreclr/tools/Common/Compiler/NativeAotNameMangler.cs	Switches string-literal mangling cache/value to Utf8String, removes old UTF-16 helper, adds replacement-char detection + hashing adjustments.
src/coreclr/tools/Common/Compiler/NameMangler.cs	Updates the abstract API to return Utf8String for mangled string names.
src/coreclr/tools/aot/ILCompiler.Compiler/Compiler/DependencyAnalysis/NodeFactory.cs	Updates UTF-8 string symbol naming to use the new Utf8String-returning mangler API.

Comments suppressed due to low confidence (1)

src/coreclr/tools/Common/Compiler/NativeAotNameMangler.cs:632

• The surrogate-special-case hash now uses MemoryMarshal.AsBytes(literal.AsSpan()), which hashes the in-memory UTF-16 byte order. The removed GetBytesFromString previously forced little-endian order, so this change can produce different mangled names on big-endian hosts. If big-endian builds are still supported, preserve a stable (little-endian) byte order when hashing UTF-16 data.

            lock (this)
            {
                _mangledStringLiterals.TryAdd(literal, mangledName);
            }