Selection constructor should accept ranges array

[Reinmar]

Cause, currently I have to do this:

const sel = new Selection();
sel.addRange( range );

Which could be simplified.

[ma2ciek]

As I talked with @Reinmar, we'll create a new static function rather than change the constructor, so there'll be no need to rewrite existing code.

Usage

const sel = Selection.createFromRanges( [ range1, range2 ] );
 // And with isLastBackward set to true:
const sel2 = Selection.createFromRanges( [ range1, range2 ], true );

Note

The Selection.createFromRanges uses the Selection#setRanges under the hood, so it's firing change event for now which is not needed I suppose.

[Reinmar]

The Selection.createFromRanges uses the Selection#setRanges under the hood, so it's firing change event for now which is not needed I suppose.

It's not needed but it changed nothing because you can't listen on it :D

As I talked with @Reinmar, we'll create a new static function rather than change the constructor, so there'll be no need to rewrite existing code.

TBH, I think I might've changed my mind. Range() does accept two positions and Position() does accept root and path. So, since range + direction is the most typical way of creating a selection I guess it could be the constructor's option.

Why I initially thought about moving it to a separate method was that I wasn't sure whether this is an identical case to Range and Position because the view selection can also be a fake one (so that's like another type of selection so why would the constructor handle just one). But... I've just completely overcomplicated this.

[ma2ciek]

Ok, it makes sense, so I'll move this code to the constructors :D

[ma2ciek]

I pushed the corrected version which still uses setRanges(). But now setRanges is reassigning _ranges and _lastRangeBackward and fires change event which is unnecessary. Maybe we can use the default parameters in the constructor function + simple range instanceof checks at the end?

[Reinmar]

Let's discuss this under ckeditor/ckeditor5-engine#918.