Smart indentation on separated elements should reuse the prior element's starting base indentation

Consider what I'd call some "ideal" smart indentation for the following code samples:

type Foo =
    | "hello"
    | "world";

{| "indentation": 0 |}

let x = 10,
    y = 2;
{| "indentation": 0 |}

let x = 10,
    y = 2,
    {| "indentation": 4 |}

let obj = {
    x: "hello" +
        "world",
    {| "indentation": 4 |}
}

class C {
    x = "hello" +
        "world";
    {| "indentation": 4 |}
}

I've specified the "expected" indentation in each of these cases; however, TypeScript today keeps indentation are the same level as the prior line which is wrong. This is especially annoying for the multi-line union case.

There is some ambiguity when one uses an ASI-assuming style though. In these cases, I would personally err on the side of expecting the start indentation.

type Foo =
| "hello"
| "world"
{| "indentation": 0 |}

```ts
let x = 10,
    y = 2
{| "indentation": 0 |}

class C {
    x = "hello" +
        "world"
    {| "indentation": 4 |}
}

Then, even for cases where ASI doesn't really apply:

let obj = {
    x: "hello" +
        "world"
    {| "indentation": 4 |}
}

I would say we should pick the start element's base (where x starts), since we'll start doing #52899 eventually.

Aucun fichier d’implémentation ni test n’est indiqué. Commencez par localiser le point d’entrée de l’indentation intelligente de TypeScript et comparez son comportement à celui des exemples d’unions séparées, de variables multilignes, d’objets et de classes ; examinez la relation avec #52899. Le travail est considéré comme terminé lorsque les exemples utilisent l’indentation de base initiale de l’élément précédent, y compris les cas d’ASI indiqués, avec une couverture de régression.