Add a generic IndexSpider class

[yolile]

The Honduras API returns a json like:

  "releases": 991869,
  "pages": 99187,
  "page": "1",
  "next": "http://www.contratacionesabiertas.gob.hn/api/v1/record/?format=json&page=2",
  "previous": null,
  "recordPackage": {...}

We can use the pages field to build the next url in a loop and make the request in parallel

[jpmckinney]

Let's also check the other spiders that inherit from LinksSpider, in case any return the total count of pages/results.

In doing so, we can create a new superclass for all spiders that can generate requests based on a total count (maybe name it IndexSpider). These include at least:

• canada_montreal
• chile_base (maybe – pagination is done in the URL path instead of the query string)
• kenya_makueni
• mexico_administracion_publica_federal
• mexico_quien_es_quien
• uganda_releases
• uk_contracts_finder

We can add a note to either the docstring or writing-spiders.rst to say that we prefer using IndexSpider instead of LinksSpider where possible, since generating all requests at once will lead to a faster collection.

[romifz]

Closed by #497