fix(snowflake): rewrite COUNT(*) FILTER to COUNT_IF [CLAUDE]

[chuenchen309]

filter_sql rewrites an aggregate FILTER (WHERE cond) into a conditional
argument by wrapping the aggregate's input in IFF(cond, input, NULL). When the
aggregate is COUNT(*), that input is a star, so the output is
COUNT(IFF(cond, *, NULL)) — a star isn't a valid IFF argument and Snowflake
rejects it.

COUNT(*) FILTER (WHERE cond) is just a count of qualifying rows, which is
exactly Snowflake's native COUNT_IF(cond), so emit that instead. Other
aggregates and COUNT(expr) / COUNT(DISTINCT ...) are untouched.

COUNT(*) FILTER (WHERE b > 0)   duckdb -> snowflake
  before: SELECT COUNT(IFF(b > 0, *, NULL)) FROM t   (invalid)
  after:  SELECT COUNT_IF(b > 0) FROM t

This completes #7884, which handled the DISTINCT / ORDER BY argument shapes
but not the star case. Added a regression test alongside the existing
FILTER-rewrite cases in test_duckdb.py; make unit and make style pass.

---

This PR was authored by an AI coding agent (Claude Code) running on this account:
the AI found the bug, ran the repro, wrote the test, and wrote this description.
The human account holder reviews every change and is accountable for it. The
verification above is real and re-runnable from the diff. If this isn't the kind
of contribution you want, say so and I'll close it.

[geooo109]

Actual the current approach works

snowflake output: SELECT COUNT(IFF(b > 0, *, NULL)) FROM t

WITH t(b) AS (
  SELECT * FROM VALUES (1), (2)
)
SELECT COUNT(IFF(b > 0, *, NULL)) FROM t;
> 2

WITH t(b, c) AS (
  SELECT * FROM VALUES (1, 'x'), (-1, 'x'), (5, 'y'), (0, 'y')
)
SELECT COUNT(IFF(b > 0, *, NULL)) FROM t;
* expands to 2 arguments and we get:
> SQL compilation error: error line 4 at position 13 too many arguments for function [IFF(T.B > 0, T.B, T.C, null)] expected 3, got 4

 

If t contains only 1 column.