feat: introduce byte, short, decimal, timestampNTZ types

[nord-erik]

adds support to natively write these spark types to neo4j:

• Byte: will convert to Integer, therefore cannot read-back as Byte
• Short: will convert to Integer, therefore cannot read-back as Short
• Decimal: will convert to String, therefore cannot read-back as Decimal
• Binary: will convert to ByteArray and will be read-back as ByteArray
• TimestampNTZType: will convert to LocalDateTime

BREAKING CHANGE:

• already existing Timestamp now converts to ZonedDateTime, not LocalDateTime.

Before, writing a Timestamp would remove the timezone information.
Now it will be normalized to UTC-time instead. If you want to retain old behaviour,
please use TimestampNTZType or be mindful when you read the stamp.

Assuming you interpret time zone when reading and writing, you should not notice.
But if you indeed intend to use LocalDateTime, this is moved and maps to TimestampNTZType

[fbiville]

question: we cannot really have null zoned datetimes in lists, can we?

[nord-erik]

I'd assume so as well. No idea what that would look like / represent like if it was the case

[fbiville]

in that case, we cannot call createArray with true. Afaik, Cypher does not support nulls in lists.

[nord-erik]

hmm... confused about this to be honest. when I did durations I added both {true, false} because timestamps already had it.

what's the consequences here to remove? I'll remove and see if all tests pass in the meantime.