ES-2276 | support level 2 query profiling

[aMahanna]

• Introduces support for setting the profile parameter in db.aql.execute to 2, which returns additional nodes information via cursor.statistics(), along with the query execution plan (newly available via cursor.plan())
• Warrants an 8.2.0 release

pip install git+https://github.com/arangodb/python-arango@ES-2276

from arango import ArangoClient

db = ArangoClient().db()

cursor_a = db.aql.execute('FOR d IN doc SORT d._key RETURN d', profile=True)
cursor_b = db.aql.execute('FOR d IN doc SORT d._key RETURN d', profile=1)
cursor_c = db.aql.execute('FOR d IN doc SORT d._key RETURN d', profile=2)

assert cursor_a.statistics().keys() == cursor_b.statistics().keys()
assert cursor_a.profile().keys() == cursor_b.profile().keys()
assert cursor_a.plan() == cursor_b.plan() == None
assert "nodes" not in cursor_a.statistics()

assert cursor_c.plan().keys() == {'nodes', 'estimatedNrItems', 'collections', 'rules', 'variables', 'isModificationQuery', 'estimatedCost'}
assert "nodes" in cursor_c.statistics()

[aMahanna]

we're starting to blur the lines between camelCase & snake_case results...

plan is a dictionary, with some fields (such as nodes or variables) being list of dictionaries, whose fields can widely vary from each other (i.e one node dictionary may not have the same fields as another node dictionary).

It becomes a slippery slope trying to convert every single field within the nodes & variables dictionaries into snake_case, so I'm opting towards just returning the camelCase fields as-is.

Similar conversation here: #321 (review)