`avg(distinct)` support

[WinkerDu]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

DataFusion does not avg(distinct ..):

> select avg(distinct a), avg(distinct b) from values(1, 2) t(a, b);
Execution error: avg(DISTINCT) aggregations are not available

Somewhat confusingly DataFusion does appear to support a single distinct avg

> select avg(distinct a) from values(1, 2) t(a, b);
+-------------------+
| avg(DISTINCT t.a) |
+-------------------+
| 1.0               |
+-------------------+
1 row(s) fetched.
Elapsed 0.015 seconds.

But it does this by rewriting effectively rewriting the query select avg(distinct a) to select avg(a) from (select distinct a ...) in the SingleDistinctToGroupBy rule

Describe the solution you'd like
Support multile distinct avg aggregates:

datafusion/datafusion/functions-aggregate/src/average.rs

Lines 116 to 118 in 2d30334

	if acc_args.is_distinct {
	return exec_err!("avg(DISTINCT) aggregations are not available");
	}

Describe alternatives you've considered
Basically we could follow the model of COUNT(distinct ...) aggregates:

In this case, there is a separate accumulator type that is implemented for count distinct

https://github.com/apache/datafusion/blob/5d08325165c1a7b32e5e35164919e83d46735e98/datafusion/functions-aggregate/src/count.rs#L234-L233

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

[qazxcdswe123]

take

[Jefffrey]

@qazxcdswe123 I see you mentioned you were too busy to continue with the related PRs: #15413 (comment)

I plan to build upon your PRs to hopefully get this issue over the line, let me know if you have any concerns or if you're able to pick this back up again 👍