Improve handling of encoding of X.520 attributes

[vcsjones]

Our managed encoder for distinguished names took a loose approach to encoding names which could result in violating the encoding rules of certain X.520 attributes. For example, the "countryName" attribute (2.5.4.6) is a PrintableString, and only a PrintableString. However, the ForceUTF8Encoding flag would result in a UTF8String, which is not permitted.

This PR brings our managed encoder to align more similarly with Windows, but makes some deviations.

1. We are now aware of X.520 attribute encoding rules. The ForceUTF8Encoding flag will only work on things that can be encoded as UTF-8. That means DirectoryString and attributes that are unknown.
2. Previously, even without ForceUTF8String, we would encode something as a UTF8String if it could not be represented as a PrintableString. Now, if you attempt to encode a PrintableString (or NumericString) with characters that are outside of the encoding rules, an exception is thrown instead of promoting it to UTF-8. This behavior matches Windows and aligns more with the specification. It is a breaking change however.

I expect few if anyone to be bothered by the stricter encoding requirements. Windows already had them, with the exception of two attributes (countryName3C and countryName3N). The breaking change doc will explain developers can use X500DistinguishedNameBuilder which offers an escape hatch and overriding of opinions on how to encode a particular component.

Fixes #109156

[vcsjones]

I'll make the breaking change doc in the next day or two.

[vcsjones]

Breaking change doc: dotnet/docs#43284