Implement translation of selected area of web page

[vitonsky]

Introduction

Related to #310

Implement feature, to let user select some area of page, then take snapshot of the area, recognize text on the image and translate text.

Use cases

Sometimes user see a pictures with text that may not be translated with DOM. Currently, for cases like this user may only say "huh" and use google app to move camera on image and see a translation. This is a not perfect, because user automatically send their data to google and experience is not smooth.

Let's implement scanner that may translate page areas, to let users translate any content in the web.