[FEATURE] Support conditional count in PPL

[penghuo]

Is your feature request related to a problem?
currently, PPL does not support conditional count. for instance, source=tbl | stats count_if(age > 18)

What solution would you like?

• PostgreSQL support, FILTER clause, for instance, select count(*) FILTER (WHERE age > 18) FROM tbl. This is a SQL-specific implementation that doesn't fit into the PPL design.
• Kusto support countif aggregation, for instance, tbl | summarize countIf(age>18). It is preferred solution.
• Splunk support eval in aggregation, for instance, source=tbl | stats count(eval(age>18)). Semantically, this approach is not ideal—eval(age > 18) returns a boolean value, and counting the boolean results would yield the same count as if no filter were applied.

What alternatives have you considered?
Using condition function, NULLIF / IF / CASE with count, and NULL value should ignored by count, for instance, source=tbl | stats count(NULLIF(age < 18))

[LantaoJin]

Seems no difference between source=tbl | stats count_if(age > 18) and source=tbl | where age > 18 | stats count()

[penghuo]

Seems no difference between source=tbl | stats count_if(age > 18) and source=tbl | where age > 18 | stats count()

count_if can help query include multiple aggregation functions, for instance, source=tbl | stats avg(age), count_if(age > 18)`