[SPARK-52428] Add UDF registration and invocation support

[rafafrdz]

Summary

• Add udf.rs module with UdfRegistration struct accessible via spark.udf()
• register_java(name, class_name, return_type) — register Java UDFs by class name
• register_java_udaf(name, class_name) — register Java UDAFs
• Add call_udf(name, args) function — calls registered UDFs (is_user_defined_function=true)
• Add call_function(name, args) function — calls any function by name (uses CallFunction proto)

// Register a Java UDF
spark.udf().register_java("my_udf", "com.example.MyUDF", DataType::String).await?;

// Use it in queries
df.select([call_udf("my_udf", vec![col("input")])]);

Test plan

• test_call_udf_expression_structure — verifies proto structure with args
• test_call_udf_no_args — verifies empty args case
• test_call_function_expression_structure — verifies CallFunction proto
• test_call_udf_is_user_defined — verifies is_user_defined_function flag differs from invoke_func
• test_register_java_udf_proto_structure — verifies Java UDF proto fields
• test_register_java_udf_return_type — verifies DataType conversion in proto
• test_register_java_udaf_proto_structure — verifies UDAF with aggregate=true
• test_register_java_udf_creates_valid_command — verifies RegisterFunction command wrapping
• cargo build passes
• cargo clippy passes
• cargo fmt -- --check passes