Reduce the number of suppressions for internal access

[qwwdfsad]

While backporting K2 commits, I've spotted one related to internal declarations.

Instead of backporting it, I decided to tide it up in the face of https://youtrack.jetbrains.com/issue/KT-60866/Phase-out-usages-of-SuppressINVISIBLEREFERENCE-INVISIBLEMEMBER-in-libraries

[qwwdfsad]

Note that it is technically a breaking change: it provide FQ name as serial name (as we do and recommend doing in the serialization) and also replaces Month (which is a copy-paste error I believe)

[ilya-g]

By what principle this FQ name is built? Is it the name of the serialized type?

[qwwdfsad]

According to the documentation:

For generated serializers, serial name is equal to the corresponding class's fully-qualified name or, if overridden, SerialName. Custom serializers should provide a unique serial name that identify both the serializable class and the serializer itself, ignoring type arguments, if they are present.

We consider this one as "the default generated serializer" in fact, thus FQN of the serialized type.
I also opened Kotlin/kotlinx.serialization#2581 to make it clear

[ilya-g]

"FQN of the serialized type" - ok, then it should be changed as there is no type with this FQN :)
Perhaps to "kotlinx.datetime.DayOfWeek"

[qwwdfsad]

I've changed it consistently across the codebase.
It also led to a format change in tests that you can observe in a separate commit.

[ilya-g]

Please exclude InlineOnly annotation removal, there's no problems with it in K2, as I was told in KT-60866

[qwwdfsad]

Done

[ilya-g]

By what principle this FQ name is built? Is it the name of the serialized type?

[qwwdfsad]

@ilya-g sorry for the delay, it got lost from my radar. PTAL