Refactor ordered-set aggregate Dataframe APIs to align with SQL

[Jefffrey]

Is your feature request related to a problem or challenge?

See following functions:

datafusion/datafusion/functions-aggregate/src/percentile_cont.rs

Lines 70 to 72 in eef1c9e

	/// Computes the exact percentile continuous of a set of numbers
	pub fn percentile_cont(order_by: Sort, percentile: Expr) -> Expr {
	let expr = order_by.expr.clone();

datafusion/datafusion/functions-aggregate/src/approx_percentile_cont.rs

Lines 55 to 61 in eef1c9e

	/// Computes the approximate percentile continuous of a set of numbers
	pub fn approx_percentile_cont(
	order_by: Sort,
	percentile: Expr,
	centroids: Option<Expr>,
	) -> Expr {
	let expr = order_by.expr.clone();

datafusion/datafusion/functions-aggregate/src/approx_percentile_cont_with_weight.rs

Lines 46 to 53 in eef1c9e

	/// Computes the approximate percentile continuous with weight of a set of numbers
	pub fn approx_percentile_cont_with_weight(
	order_by: Sort,
	weight: Expr,
	percentile: Expr,
	centroids: Option<Expr>,
	) -> Expr {
	let expr = order_by.expr.clone();

Issues are they accept a Sort which extracts the expression to calculate for, however this allows specifying null_order which is meaningless as they will always ignore nulls in their calculations.

Describe the solution you'd like

Consider refactoring to something like:

pub fn percentile_cont(expr: Expr, percentile: Expr, asc: bool) -> Expr

Describe alternatives you've considered

Don't do this

Additional context

See: #17805 (comment)