[Ballista] support datafusion built_in UDAF work in ballista cluster

[Ted-Jiang]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.
support datafusion built_in UDAF work in ballista cluster, It will make ballista more like a high-performance OLAP system for
production.

❯ select APPROX_DISTINCT(14) from table1;
[2022-03-11T03:27:22Z INFO  ballista_core::execution_plans::distributed_query] Connecting to Ballista scheduler at http://localhost:50050
Execution("Status { code: Internal, message: \"Job Gs7JJH3 failed due to General error: 
Could not save stage plan: Not implemented: Aggregate function not supported: ApproxDistinct { name: \\\"APPROXDISTINCT(Int64(14))\\\", input_data_type: Int64, expr: Literal { value: Int64(14) } }\",
 metadata: MetadataMap { headers: {\"content-type\": \"application/grpc\", \"date\": \"Fri, 11 Mar 2022 03:27:22 GMT\"} }, source: None }")
❯

Describe the solution you'd like
A clear and concise description of what you want to happen.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

[Ted-Jiang]

work in client without ballista

select APPROX_DISTINCT("14") from table1;
+---------------------------+
| APPROXDISTINCT(table1.14) |
+---------------------------+
| 55                        |
+---------------------------+
1 row in set. Query took 0.015 seconds.

[jiangzhx]

FYI，It seems to be the same problem
#1881

[Ted-Jiang]

FYI，It seems to be the same problem #1881

IMO, this PR not work for built in UDAF.

[Ted-Jiang]

@houqp plz assign this to me. We need this in our situation.

[liukun4515]

I think many functions are missed in the ballista but implemented in the datafusion.
@Ted-Jiang
You can give a plan to find them and fix them through the community.

[liukun4515]

@Ted-Jiang
Do you mean the same thing with this issue https://github.com/apache/arrow-datafusion/issues/1988？
We should make it clear database system just have two type of functions, one is the built-in function and the other is UDF/UDAF.
You can refer to https://spark.apache.org/docs/latest/sql-ref-functions.html and get some knowledge about that.
#1881 just want to resolve the issue of adding a new UDF/UDAF.