Add support for data with schema with non-struct top level

[ikbalkaya]

We need to consider converting data with schema where schema.type != STRUCT at top level. Currently messages with top level schemas other than STRUCT won't be JSONified. BYTES and STRINGs will be passed as is as a data to Ably message. Other top level types are not supported and connector will throw an exception. While this seems enough good for most cases, we should consider supporting all top level Kafka connect schemas.

I am not sure how likely it is a top level schema could be a primitive type, but I think we should still address it. For complex types we already support BYTES and STRUCT, We just need to address ARRAY and MAP types.

This issue is created from conversation below

If I understand correctly, this will fail if valueSchema != null and valueSchema != STRUCT, i.e. there is a schema set, but it's not a Struct type, such as Array, String, Integer, etc, as we'll still be passing a Connect data type to ably-java, which it won't know how to process?

Have we made a deliberate decision to only support Struct as the top-level schema? If so, we should probably add a more explicit error, else it'll be raised as a bug should somebody try to send an array or similar. I'm not sure it's really necessary to require this? Are not 1, "foo", 3.145, [1, 2, 3] etc all valid JSON values? The docs suggest we at least support arrays.

If others agree, I'm happy with us just raising a ticket to add support for other top-level types later rather than delaying this PR.

Originally posted by jaley in #73 (comment)

┆Issue is synchronized with this Jira Uncategorised by Unito