Implement Spark’s FunctionCatalog for Existing Transformations

[kbendick]

We need to implement Spark’s FunctionCatalog so that we can use the partition transformation functions in queries.

This allows for using the partition transforms on non-partition columns in generated code.

This is necessary in order to write Catalyst rules which will pass bucket So that storage partitioned joins (aka bucketed joins) can be implemented.

See also:

• FunctionCatalog
• ScalarFunction interface, which has practical description of what is needed for codegeneration and the benefits of it.

The functions we have that are likely highest priority:

• truncate
• bucket
• zorder
• date transformations

[kbendick]

This will allow us to make use of Spark’s storage partitioned joins (aka bucket joins which is one subset of possible join optimizations of transformed columns) https://issues.apache.org/jira/browse/SPARK-37166

[kbendick]

This relates to #430