Idea: json menus 🍜

Reading restaurant menus with a reading device is often a nuisance.
Text gets scrambled up as soon as the font styles and spacing get a little bit fancy.

On the other hand, restaurants often put their menus online, so the idea is to test and provide a speaker solution for online menus.

0. try PDF speaker on a few menus
1. provide a recipe to JSONify the dishes
2. fields include:
* a list of special ingredients and hints (garlic, spicy, wheat, nuts)
* a list of general keywords (special dishes to be separated into whether including chicken, cheese)
* price
* name, number
3. some grep functionality, type a token to find examples of it.

No files, tests, or entry points are named. Start by trying PDF speaker output on several online restaurant menus, then define the JSON dish fields and search behavior; done means an agreed recipe and a working speaker-oriented menu representation.

Written by the indexing model from the issue text.