[MODEL] Wrong retrieval of records corresponding to indexed documents when querying multiple models.

[miguelff]

This has been shown in a production deployment of elasticsearch-model v0.17.0

In this commit some refactorings where made to the adapter for multiple records.

One of the refactorings was a reduction of the cyclomatic complexity (and renaming of __type_records):

This change drove to the following

• A user searches across several models. Along the search execution the code reaches the point above.
• A certain klass (say for instance an ActiveRecord model) wont' match any of the cases on the case/when statement, entering in the default clause.
• Then (line 60 on the right side) find will be invoked over klass which is an ActiveRecord, but unluckily ActiveRecord will respond to the message.
• Due to eventual consistency, some of the records returned by ES, are no longer present in the DB.
• So in this case, klass.find(ids) will raise an exception as no all the ids where found in the DB. 💥

I'm preparing a fix for this.