feat: Expose fixed-array inner product in SQL

[0guban0v]

Closes #28928

Adds ARRAY_INNER_PRODUCT(lhs, rhs) and ARRAY_DOT_PRODUCT as SQL entry points to native fixed-size Array dot path introduced in #28504 and extended across integer dtypes in #28829.

Lowering deliberately ends at Expr::arr().dot() instead of rebuilding (lhs * rhs).arr.sum(). SQL inherits supertype coercion, output dtype, integer wrapping, null behavior, execution-engine behavior, and allocation improvements from #28980 while avoiding intermediate product Array.

Direct SQL array literals passed to this function are represented as scalar fixed-size Arrays. Everywhere else, SQL array literals remain List-backed, avoiding global coercion changes. Query-vector form matches native one-row broadcasting, and #29000 can preserve scalar through dispatch and broadcast result without SQL-specific follow-up.

Coverage includes both aliases, mixed inner dtypes, literal syntax and argument position, all-null and constant literals, projection-height broadcasting, native plan shape, non-Array rejection, arity, and width validation.

Example:

import polars as pl

dtype = pl.Array(pl.Float64, 2)
df = pl.DataFrame(
    {
        "lhs": [[1.0, 2.0], [3.0, 4.0]],
        "rhs": [[10.0, 20.0], [30.0, 40.0]],
    },
    schema={"lhs": dtype, "rhs": dtype},
)

result = df.sql("""
SELECT
  ARRAY_INNER_PRODUCT(lhs, rhs) AS rowwise_dot,
  ARRAY_DOT_PRODUCT(lhs, [10.0, 20.0]) AS query_dot
FROM self
""")

print(result)

Output:

shape: (2, 2)
┌─────────────┬───────────┐
│ rowwise_dot ┆ query_dot │
│ ---         ┆ ---       │
│ f64         ┆ f64       │
╞═════════════╪═══════════╡
│ 50.0        ┆ 50.0      │
│ 250.0       ┆ 110.0     │
└─────────────┴───────────┘

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.47%. Comparing base (5fb8434) to head (ea68b05).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #29033      +/-   ##
==========================================
+ Coverage   81.40%   81.47%   +0.06%     
==========================================
  Files        1888     1888              
  Lines      267825   267855      +30     
  Branches     3061     3061              
==========================================
+ Hits       218012   218222     +210     
+ Misses      49033    48854     -179     
+ Partials      780      779       -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.