[CALCITE-7806] Avoid resolving generated CAST calls during array validation

[FrankChen021]

Fixes CALCITE-7806.

Why

ARRAY and MAP validation creates CAST calls when operands must be adjusted to a common element type. These calls already contain Calcite's built-in CAST operator, but a later deriveType resolves the same operator by name for every generated CAST, allocating temporary operand and argument-type lists along the way.

In Druid's string-IN planning benchmark, avoiding that repeated lookup reduced allocation by 19.95% at 100,000 literals and 20.48% at 1,000,000 literals. Timing confidence intervals overlapped, so no latency improvement is claimed.

What

Before this change:

Create built-in CAST node
  -> no validated type recorded
  -> later validator.deriveType(cast)
  -> build operand and argument-type lists
  -> resolve "CAST" through the operator table
  -> validate operands and store the type

With this change:

Create built-in CAST node
  -> validate through cast.getOperator()
  -> register the validated type immediately
  -> later validator.deriveType(cast)
  -> return the cached type

This preserves CAST operand validation and return-type inference while avoiding resolution of an operator that is already known.

Verification

• SqlValidatorTest verifies that a generated CAST is registered and that a later deriveType performs no operator-table lookup.
• ./gradlew :core:test --tests org.apache.calcite.test.SqlValidatorTest :core:autostyleJavaCheck :core:checkstyleTest (596 completed, 0 failed, 7 skipped)
• Druid InPlanningBenchmark.queryStringInSqlPlanOnly with -prof gc

Druid benchmark results

String literals	Baseline allocation	PR allocation	Reduction
100,000	1,448.09 MiB/op	1,159.19 MiB/op	288.90 MiB/op (19.95%)
1,000,000	14,653.13 MiB/op	11,651.62 MiB/op	3,001.52 MiB/op (20.48%)

Configuration: 2 forks, 2 one-second warmup iterations, and 5 one-second measurement iterations. Allocation is cumulative bytes per operation, not retained or peak heap.

The performance measurement currently comes from Druid; a Calcite-local ubenchmark is not yet included.

Scope

This change applies only to CAST calls generated while adjusting ARRAY and MAP constructor operands. It does not change CAST semantics or user-written CAST validation.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[julianhyde]

-1 slop

[mihaibudiu]

You seem to do strictly more work than before.
This looks correct, but does not seem more efficient.