Support skewness(x) in Aggregation function

[dharanad]

Which issue does this PR close?

Part of #12254
Closes #12249

Implemented by referring to https://github.com/duckdb/duckdb/blob/main/src/core_functions/aggregate/distributive/skew.cpp

Rationale for this change

What changes are included in this PR?

Added skewness UDAF.
Ref: https://duckdb.org/docs/sql/functions/aggregates#skewnessx

Are these changes tested?

Are there any user-facing changes?

[dharanad]

Planning to add more logic tests.

[2010YOUY01]

Planning to add more logic tests.

I think we can port tests from duckdb https://github.com/duckdb/duckdb/blob/main/test/sql/aggregate/aggregates/test_skewness.test

[comphead]

Thanks @dharanad the PR brings a lot of value.
Would you mind resolving conflicts?

[comphead]

I just realized we dont have any examples in aggregate_functions.ms

[comphead]

Basically I'm feeling we have too much of user .md files, duplicating the functions description and poorly connected to each other which confuses a lot.

@alamb WDYT about having a single file for all SQL functions/scalars/expressions like in https://spark.apache.org/docs/latest/api/sql/index.html with description, examples

Btw Spark holds an internal meta repository that allows such doc to be regenerated from the code, which can be next step

[alamb]

Btw Spark holds an internal meta repository that allows such doc to be regenerated from the code, which can be next step

I agree

What I actually think would be ideal is exactly as @comphead describes and put all the documentation about the functions in the code, and then have a post processing step that auto generates the documentation

This is how the configuration setting documentation is created. Here is the script that does it:
https://github.com/apache/datafusion/blob/main/dev/update_config_docs.sh

BTW I think this is the direction @findepi is heading with #12266. I also think GlareDB did something similar, but I can't find the link now.

[alamb]

I also wonder if we should consider where to draw the line on features to include (i.e. should we include skewness by default in the core?) -- especially as all aggregate functions use the same API, this is the kind of feature we could very easily keep outside -- either in its own crate or even own repo -- and then have other code integrate it in -- e.g. #11979