Use UTF-8 constants in serialization/deserialization code

[AlexanderSher]

https://github.com/Azure/azure-sdk-for-net/ repo is updated to C# 11, so now we can use UTF-8 string literals. As @chunyu3 has found, utf8 spans give us 30% performance boost in serialization/deserialization comparing to strings.

[archerzz]

LGTM. Just wonder if we could update CodeWriter to append u8 to all string literals.

[AlexanderSher]

LGTM. Just wonder if we could update CodeWriter to append u8 to all string literals.

No. We can append u8 only where there is matching method signature. E.g., we can't create diagnostic scope with ReadOnlySpan<byte> instead of string. Same with headers - there is no overload of System.Net.Http.Headers.HttpHeaders.TryAddWithoutValidation that accepts ReadOnlySpan<byte>.