Support aggregations across constants

[costin]

Description

In ESQL it is possible to have aggregations working on a constant as long as it comes from a source, however pushing it into the aggregation fails.

This works: row a = 1 | stats m = min(a) but the query stats m = min(1) does not.
We've been looking at adding a dedicated rule for optimizing this ( #100182) however the issue is rather related to folding aggregations.

A somewhat variant of this is folding the aggregation when the argument is the group key itself:
stats mi = min(k), ma = max(k), c = count(k), a = avg(k) by k which is the same as stats by k | keep k as mi, k as ma, k as c, k as a, k

• Implement for aggs where agg(const) can be expressed using a multi-value function as mv_...(const) ESQL: Sum, Min, Max and Avg of constants #105454 ESQL: median, count and count_distinct over constants #107414
• Remove corresponding awaitsFixes from tests, and enable commented-out code
• Implement for aggs without corresponding multi-value functions (PERCENTILE, MEDIAN_ABSOLUTE_DEVIATION ST_CENTROID_AGG)
• Enable PropagateEvalFoldables for aggregates

[elasticsearchmachine]

Pinging @elastic/es-ql (Team:QL)

[elasticsearchmachine]

Pinging @elastic/elasticsearch-esql (:Query Languages/ES|QL)

[alex-spies]

I think this was closed on accident; just checked and running ROW a = 1 | stats m = min(1) still returns an EsqlIllegalArgumentException (500). Reopening.

[elasticsearchmachine]

Pinging @elastic/es-analytics-geo (Team:Analytics)

[alex-spies]

The second variant mentioned in the description cannot be always folded due to multi-values:

POST /testidx/_doc
{"k": [0,1,2,3]}

POST /_query
from testidx | stats mi = min(k),  ma = max(k), c = count(k), a = avg(k) by k

->
      mi       |      ma       |       c       |       a       |   k
---------------+---------------+---------------+---------------+---------------
0              |3              |4              |1.5            |0              
0              |3              |4              |1.5            |1              
0              |3              |4              |1.5            |2              
0              |3              |4              |1.5            |3 

[alex-spies]

Let's also fix STATS COUNT([1,2,3]) as part of this issue (currently - wrongly - treated the same as COUNT(*)) - this should be replaced by STATS MV_COUNT([1,2,3]) * COUNT(*).

[elasticsearchmachine]

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