feat: support Spark format_string expression

[n0r0shi]

Summary

• Wire format_string from the datafusion-spark crate (FormatStringFunc) to Comet
• Register in jni_api.rs and add serde mapping via CometScalarFunction in QueryPlanSerde.scala

Test plan

• New test in CometStringExpressionSuite — passes
• Spotless formatting check passes
• Rust native build passes

[andygrove]

Clean serde wiring. The CometScalarFunction approach and UDF registration look correct.

For the tests, it would be a good fit to use the SQL file framework instead of CometStringExpressionSuite. There are already many string expression SQL files in spark/src/test/resources/sql-tests/expressions/string/. A format_string.sql file there would automatically compare Spark and Comet results.

It might also be worth adding a test with %f for floating-point formatting, since Spark uses java.util.Formatter with US locale and there could be subtle differences in how Rust formats floats. Something like format_string('%.2f', 3.14159) would exercise that path.