[menato]

It looks like Yandex intentions is similar to what JavaScript did to browsers.

Initially browsers could just ask for a static content. Later some interactivity was added and sites changed from beeing a [content] entity to a [content, on-action] entity.

Here we also have: initially sites were static for search engine, but now there is an interactivity support, which should split monolith entity of site.

[winter_blue]

Most search engines (and Yandex too I assume), crawl the static web. So if you have a web page whose content is auto-generated using Javscript, it likely won't be indexed by many search engines. Google has been working on this for a while, but it's definitely a very hard problem (tm) to solve (and computationally expensive too).

[menato]

It's not exactly about parsing JS output: JS output is something which is dedicated for user of webpage.

Ilands interface of site is something dedicated to search engine from the origin and helps to create a new layer between search engine and site.