Update the format of the comparison function output

[tcezard]

In this issue we discussed changing the format of the comparison function to address future update. The main changes are:

• rename "arrays" to "attributes"
• rename "elements" to "array_elements"
• Within array_elements, explode the total to get the count of level2 elements per attribute

Before

{
  "digests": {
    "a": "8djrpzjdbsoeghbadoadq.", 
    "b": "psuhfbsjwttzaywhdjsid"
  },
  "arrays": {
    "a_only": [],
    "b_only": [], 
    "a_and_b": ["names", "lengths", "sequences"],
  "elements": {
    "total":{
       "a":3,
       "b":3,
    },
    "a_and_b": {
       "names":2,
       "lengths":2,
       "sequences":2
    }
    "a_and_b_same_order": {
       "names":true,
       "lengths":true,
       "sequences":true
    }
  }
}

After

{
  "digests": {
    "a": "8djrpzjdbsoeghbadoadq.", 
    "b": "psuhfbsjwttzaywhdjsid"
  },
  "attributes": {
    "a_only": [],
    "b_only": [], 
    "a_and_b": ["names", "lengths", "sequences"],
  "array_elements": {
    "a":{
       "names":3,
       "lengths":3,
       "sequences":3
    },
    "b": {
       "names":3,
       "lengths":3,
       "sequences":3
    },
    "a_and_b": {
       "names":2,
       "lengths":2,
       "sequences":2
    },
    "a_and_b_same_order": {
       "names":true,
       "lengths":true,
       "sequences":true
    }
  }
}

[waterflow80]

@tcezard
For the array_elements (a and b), should we consider all the attributes that a (or b) has. i.e the attributes proper to a (or b) and the the common ones too.

For example, if a has a unique attribute called new_attribute, should we have the output below:

  "digests": {
    "a": "8djrpzjdbsoeghbadoadq.", 
    "b": "psuhfbsjwttzaywhdjsid"
  },
  "attributes": {
    "a_only": ["new_attribute"],
    "b_only": [], 
    "a_and_b": ["names", "lengths", "sequences"],
  "array_elements": {
    "a":{
       "names":3,
       "lengths":3,
       "sequences":3,
       "new_attribute": 3
    },
    "b": {
       "names":3,
       "lengths":3,
       "sequences":3
    },
    "a_and_b": {
       "names":2,
       "lengths":2,
       "sequences":2
    },
    "a_and_b_same_order": {
       "names":true,
       "lengths":true,
       "sequences":true
    }
  }
} 

[tcezard]

Slight update to this in the associated issue in this comment. I've formalised the decision to ensure it stabilises in this ADR
At this point the only change is that attribute a, b, and ,a_and_b are replaced with a_count, b_count, and a_and_b_count