Core: Add CountNull Aggregation Support

[jackylee-ch]

We’re using Aggregate to generate per-column statistics from data files—metrics like value counts, min/max, and null counts—but discovered that null counts (CountNull) aren’t currently supported. This PR adds null count support for data files. It also enables any other consumer to easily collect null counts moving forward.

[jackylee-ch]

@huaxingao PTAL

[LuciferYang]

Could you help review this pr when you have time? @huaxingao Thanks ~

[kevinjqliu]

should we also add to SparkAggregates?

https://grep.app/search?f.path=spark%2F&f.repo=apache%2Ficeberg&q=Operation.COUNT_STAR

[kevinjqliu]

btw CountNonNull exists https://github.com/apache/iceberg/blob/main/api/src/main/java/org/apache/iceberg/expressions/CountNonNull.java

[huaxingao]

should we also add to SparkAggregates

I believe CountNull is not supported by Spark because it's not standard SQL, which is why I didn't add it at first; However, it's very useful for data-file statistics, so I think it's reasonable to add the support.

[kevinjqliu]

i wonder if we need to update the docs too

iceberg/api/src/main/java/org/apache/iceberg/expressions/Aggregate.java

Lines 21 to 24 in 173d4c3

	/**
	* The aggregate functions that can be pushed and evaluated in Iceberg. Currently only three
	* aggregate functions Max, Min and Count are supported.
	*/

we have since added CountStar, CountNonNull, and now CountNull

[kevinjqliu]

i wonder if this is consider a "spec change" or at least clarification.

from https://iceberg.apache.org/spec/#data-file-fields,
i see value_counts, null_value_counts, and nan_value_counts

[huaxingao]

In the original java doc Count covers both Count(*) and Count(field), but now since we also support CountNull, lets update the java doc and spell out each of them. Something like this:

 /** 
  * The aggregate functions that can be evaluated in Iceberg. Supported aggregates include
  * Min(field), Max(field), Count(*), Count(field) and CountNull(field)
  */ 

[huaxingao]

LGTM

[huaxingao]

Thanks @jackylee-ch for the PR! Thanks @kevinjqliu for the review!

[jackylee-ch]

Thanks for your preview. @huaxingao @kevinjqliu