Fix strpos argument order and add tests

[DAlperin]

Motivation

• Ensure strpos follows PostgreSQL argument order (string, substring) instead of the reversed order used previously.
• Provide a dedicated implementation and dispatch path so strpos behaves consistently with position.
• Add sqllogictest coverage to prevent regressions in argument order and behavior.

Description

• Add a Strpos variant to the BinaryFunc enum and wire it into the binary function dispatch and related trait methods (eval, output_type, propagates_nulls, introduces_nulls, is_infix_op, negate, could_error, is_monotone, and fmt).
• Implement a new Rust SQL function strpos(string: &str, substring: &str) that delegates to the existing position implementation with the arguments swapped (position(substring, string)).
• Route the SQL builtin strpos to BinaryFunc::from(func::Strpos) in PG_CATALOG_BUILTINS and add sqllogictest entries in test/sqllogictest/string.slt covering strpos behavior and examples.

Testing

• Added sqllogictest cases in test/sqllogictest/string.slt exercising SELECT strpos(strcol2, strcol1) and SELECT strpos('high', 'ig'), but no automated test suite was executed in this change.
• No additional automated tests were run; the new sqllogictest entries provide coverage to be executed by the existing test runner.
• Local compile or runtime tests were not performed as part of this change.

---

Codex Task

[petrosagg]

instead of the reversed order used previously.

I don't see any modified code in the PR, where was the strpos implementation that was used previously?

[antiguru]

Looks good, but please rewrite the PR title/description :)

[antiguru]

Suggested change

	#[sqlfunc(propagates_nulls = true)]
	#[sqlfunc]

propagates_nulls is implied from arguments that cannot represent NULL.

[antiguru]

Please add a doc comment to the function (not used right now, but I hope it'll be eventually.