Feat/bulk scoring

[rhanka]

1. move to mix "should":

• blocking on a reduce scope ("should" with minimum_should_match) by defaut on name and date (at leas one should match)
• other fields are only should (but envolved in scoring)

2. normalize score using :

• first and last name (handling switch)
• location (city, citycode, depcode, country)
• date (handling range match)
• sex

use js-levenshtein
handles pruning for faster scoring (date then name then location)

base on top 10 ES results.

2. concatenante source CSV to have complete result, using same separator as input

[cristianpb]

👌