Fix is_incremental() macro

[jtcohen6]

The default implementation of is_incremental() bundles the following four conditions:

relation is not none
and relation.type == 'table'
and model.config.materialized == 'incremental'
and not flags.FULL_REFRESH

The second condition is always false on Spark because relation objects don't naturally have a value for type. Instead, relation type is handled by the spark_get_relation_type macro, which infers it from

show tblproperties ... ('view.default.database')

There are a few potential fixes:

• Reimplement is_incremental() for dbt-spark
• Update the relation object with its type as part of the incremental materialization, since we're already calling spark_get_relation_type here
• At the beginning of runs, cache all info on dbt relations, including relation types (similar to get_catalog, although this takes a while to run)