feat: spark udf array shuffle

[chenkovsky]

Which issue does this PR close?

Rationale for this change

support shuffle udf

What changes are included in this PR?

support shuffle udf

Are these changes tested?

UT

Are there any user-facing changes?

No

[Jefffrey]

Suggested change

	signature: Signature::any(1, Volatility::Volatile),
	signature: Signature::arrays(1, None, Volatility::Volatile),

Example:

datafusion/datafusion/functions-nested/src/empty.rs

Lines 72 to 79 in 44cd972

	impl ArrayEmpty {
	pub fn new() -> Self {
	Self {
	signature: Signature::arrays(1, None, Volatility::Immutable),
	aliases: vec!["array_empty".to_string(), "list_empty".to_string()],
	}
	}
	}

(using arrays() instead of array() to avoid the coercion from FixedSizeList to List)

Although, looking at the Spark doc it says it accepts an optional seed argument; do we need to include that here?

[Jefffrey]

I think if we implement with a seed argument we can have deterministic tests for shuffle, without running it through sort or relying on the shuffled permutation not being equal to the sorted version

[Jefffrey]

Suggested change

	fn general_array_shuffle<O: OffsetSizeTrait + TryFrom<i64>>(
	fn general_array_shuffle<O: OffsetSizeTrait>(