ReadOnlyReactiveProperty<T>.ToReadOnlyReactiveProperty() has dangerously different (disposability) behavior from ReactivePropertyExtensions.ToReadOnlyReactiveProperty()

[wizardness]

Strongly related to #263
Reproduced on R3 1.3.0 (.NET 10)

The case that bit me (no shade, I love this library):

var rp = new ReactiveProperty<int>(1);
var view = rp.ToReadOnlyReactiveProperty(); // instance method returns self
view.Dispose();
rp.Value = 5;        // throws ObjectDisposedException

More generally, code that looks the same requires different disposal handling:

var rp = new ReactiveProperty<int>(1);
ReadOnlyReactiveProperty<int> asReadOnly = rp;   // cast, same object
Observable<int> asObservable = rp;               // cast, same object

ReferenceEquals(rp, rp.ToReadOnlyReactiveProperty());            // True   - instance method
ReferenceEquals(rp, asReadOnly.ToReadOnlyReactiveProperty());    // True   - instance method
ReferenceEquals(rp, asObservable.ToReadOnlyReactiveProperty());  // False  - extension method

Passing an argument also flips which method is called:

ReferenceEquals(rp, rp.ToReadOnlyReactiveProperty());     // True  - instance method (no params)
ReferenceEquals(rp, rp.ToReadOnlyReactiveProperty(99));   // False - extension method (one param)