fix(optimizer): keep projections referenced by SORT BY / DISTRIBUTE BY / CLUSTER BY

[fivetran-kwoodbeck]

pushdown_projections kept projections referenced by ORDER BY, but Spark's SORT BY, DISTRIBUTE BY and CLUSTER BY can reference select aliases too and were not checked, so their projections were pruned. Treat all three clauses as output references, hoisting them onto the exp.SetOperation when they trail a set operation so both arms keep the projection.

-- input (spark)
SELECT t.a FROM (SELECT a, b + c AS s FROM x SORT BY s) t

-- old output: alias `s` no longer exists -> UNRESOLVED_COLUMN
SELECT t.a AS a FROM (SELECT x.a AS a FROM x AS x SORT BY s) AS t

-- fixed output
SELECT t.a AS a FROM (SELECT x.a AS a, x.b + x.c AS s FROM x AS x SORT BY s) AS t

[github-actions]

SQLGlot Integration Test Results

✅ All tests passed

Comparing:

• this branch (sqlglot:fix/pushdown-projections-spark-sort-distribute-cluster @ sqlglot e3fce8b)
• baseline (main @ sqlglot d7dd2cd)

Overall

main: 182937 total, 163862 passed (pass rate: 89.6%)

sqlglot:fix/pushdown-projections-spark-sort-distribute-cluster: 182937 total, 163862 passed (pass rate: 89.6%)

Transitions:
No change

✅ All tests passed

[georgesittas]

LGTM once final concern(s) are addressed. @geooo109 wanna take another look?