[Kernel] Fixed potential float -> double casting issue and aligned IN expression with comparator "=" implementation

[Zaylinux]

Which Delta project/connector is this regarding?

• Spark
• Standalone
• Flink
• [ X ] Kernel
• Other (fill in here)

Description

Resolves #5227

• Investigated float-to-double precision issue in ImplicitCastExpression

• Refactor validateAndTransform method to use implicit casting

• Simplify value comparison logic in InColumnVector

How was this patch tested?

New UTs and validated existing UTs

Does this PR introduce any user-facing changes?

No

[huan233usc]

Why do we want to remove this file?

[huan233usc]

Can we try to move the utils defined in

delta/kernel/kernel-defaults/src/main/java/io/delta/kernel/defaults/internal/expressions/ImplicitCastExpression.java

Lines 102 to 126 in 73af97a

	private static final Map<String, List<String>> UP_CASTABLE_TYPE_TABLE =
	unmodifiableMap(
	new HashMap<String, List<String>>() {
	{
	this.put("byte", Arrays.asList("short", "integer", "long", "float", "double"));
	this.put("short", Arrays.asList("integer", "long", "float", "double"));
	this.put("integer", Arrays.asList("long", "float", "double"));
	this.put("long", Arrays.asList("float", "double"));
	this.put("float", Arrays.asList("double"));
	}
	});
	/**
	* Utility method which returns whether the given {@code from} type can be cast to {@code to}
	* type.
	*/
	static boolean canCastTo(DataType from, DataType to) {
	// TODO: The type name should be a first class method on `DataType` instead of getting it
	// using the `toString`.
	String fromStr = from.toString();
	String toStr = to.toString();
	return UP_CASTABLE_TYPE_TABLE.containsKey(fromStr)
	&& UP_CASTABLE_TYPE_TABLE.get(fromStr).contains(toStr);
	}

to implement this helper?

This method is good but want to see if we could just re-use code

[huan233usc]

Let's do the collation validation first, casting to common type is our implementation that may not be related to the user provided expression