glaze silently has different JSON schemas for parsing and serializaition due to partial parsing

[CoinFuMasterShifu]

I have detected differences between input and output JSON schemas:
Output schemas

• objects should have the required: [ <required keys list>] property because after serialization they are guarranteed to exist.
• arrays that come from tuples should also have the minItems set to the same value of maxItems property because the tuple will always serialize with full length.

Input schemas

• My understanding of the library is not deep enough but I guess due to partial parsing input schemas for objects do not have the required property.
• In a recent pull jsonchema prefixItems for tuples and arrays #2461 you added the maxItems for tuple parsing but did not include the minElements because glaze supports partial parsing (I don't find that anymore but I think I have read it)

I think we should make this difference explicit and pass an option to whether we want the input or output JSON schema of a C++ struct because strictly speaking they are different since output has other guarrantees than input has requirements.

[CoinFuMasterShifu]

The way I observed it as a problem is that I tried to use AI to document optional keys and it told me every key is optional because there are no required properties in the schema that gurrantee existence of keys. I checked the JSON schema documentation and indeed, required fields have to be listed in the required array of objects. And my interpretation was for JSON schema of output data, the required means guarranteed to exist, while not required means that the field may be misssing from the output.