Custom deserializer and empty string treatment

In our scenario, we have an API that for a field reports an empty String. The empty String should be deserialized to an enum value that represents the scenario of emptry String: "" -> ConstaintType.NONE (yes, I know this is weird).

I noticed that it however did not end up in our custom deserializer that handles this mapping. The reason for that is that JsonNullableDeserializer#deserialize(JsonParser, DeserializationContext) catches the scenario for empty string and maps it to JsonNullable.undefined(). This logic currently happens if the deserializer is not a String deserializer (there is an open PR that extends it to CharSequence https://github.com/OpenAPITools/jackson-databind-nullable/pull/45).

In my opinion, it is being a bit too smart here. I would expect that it always delegates to super; or at least only return JsonNullable.undefined for the standard Java types and delegate to super.deserialize for the custom user defined Java classes? At minimum, it would be nice if we could easily override this behaviour?

Start by reading JsonNullableDeserializer#deserialize(JsonParser, DeserializationContext) and the empty-string handling described in the issue. Compare the current behavior with the custom enum-deserializer scenario and the proposed CharSequence change in PR #45. Done means agreeing on and validating a clear policy for empty strings reaching custom deserializers, including an override or delegation path if required.