FR: `capture-html` , but with `eww-readable`

As always, thanks for the useful package!

### Feature Request / Proposal

I propose to add a capture protocol that behaves like `capture-html` currently does, but (if nothing is selected), sends the whole page's html (i.e., `document.body`) to emacs and uses `eww-readable` to extract the interesting html (before converting it with pandoc). To basically a mix of the two existing protocols.

### Use case

- Let's say I want to capture a webpage into org. If I only want to capture selected part(s) of the website, then `capture-html` works perfectly.
- On the other hand, if I want to capture the full webpage, and want to use the power of `eww-readable`, but want to capture the content I have already open, this is currently not possible. (Or I did misunderstand something.)

### Background

What I mean: If I understand things correctly,

`org-protocol-capture-html--with-pandoc` (used by the `capture-html` protocol)
1) user selects some text in browser, clicks on bookmarklet
2) selected text is send to emacs via org-protocol
3) function converts the html (coming from the browser) to org

`org-protocol-capture-html--capture-eww-readable` (used by the `capture-eww-readable` protocol`)
1) user clicks on bookmarklet (which does not send selected text to emacs)
2) function uses `org-protocol-capture-html--url-html` to _download html_ directly
3) function then converts the html (downloaded by curl) to org

The difference being that in `capture-html` the html is retrieved by the browser, while in `capture-eww-readable` it is retrieved by emacs/curl.

This makes a difference when the content being captured is, e.g., a page behind a paywall, or dynamically generated content. It is also a difference when the text is inserted into the DOM by some javascript which loads the content after the page itself is loaded.