ESQL: per aggregation filtering

[costin]

Description

Allow per-aggregation filtering:

stats ca = count(a) FILTER a > 0, cb = count(b) FILTER b > 0, c = count(), d = count(d) FILTER e > 10

P.S. There's a discussion whether to reuse WHERE or take the SQL approach and use FILTER.

[elasticsearchmachine]

Pinging @elastic/es-analytical-engine (Team:Analytics)

[dgieselaar]

An alternative suggestion for syntax (not sure what the cons are):

stats ca = count(FILTER a > 0), cb = count(FILTER b > 0), c = count(), d = count(FILTER e > 10)

[philippkahr]

Can we stick to a single term for "filtering", which is currently where? as long as where has feature parity when used inside of an aggregation?

I also prefer Darios version of writing the aggregation, as from a readability standpoint, I know where the condition is applied to and I do not need to scan for the next , to know where the filter ends.

stats ca = count(WHERE a > 0), cb = count(WHERE b > 0), c = count(), d = count(WHERE e > 10)