Speedup query to do ApiKey grants

[amiedes]

Related: https://app.clubhouse.io/cartoteam/story/117080/performance-issues-with-datasets-r-scope

What does this PR do?

Previously, when generating the GRANTs associated to an API key, we were doing a query similar to the following to retrieve the sequences associated with each of the tables the API key had access to:

SELECT
  n.nspname, c.relname
FROM
  pg_depend d
  JOIN pg_class c ON d.objid = c.oid
  JOIN pg_namespace n ON c.relnamespace = n.oid
WHERE
  d.refobjsubid > 0 AND
  d.classid = 'pg_class'::regclass AND
  c.relkind = 'S'::"char" AND
  d.refobjid IN (
    (QUOTE_IDENT('amiedes') || '.' || QUOTE_IDENT('africa_adm0'))::regclass
  );

For an API key with access to 150 tables, this query took ~200ms. This means for the 150 tables it was around 35 seconds.

This PR combines all the queries to retrieve the list of sequences to give the GRANTs in a single query, so it is faster:

SELECT
  n.nspname, c.relname
FROM
  pg_depend d
  JOIN pg_class c ON d.objid = c.oid
  JOIN pg_namespace n ON c.relnamespace = n.oid
WHERE
  d.refobjsubid > 0 AND
  d.classid = 'pg_class'::regclass AND
  c.relkind = 'S'::"char" AND
  d.refobjid IN (
    (QUOTE_IDENT('amiedes') || '.' || QUOTE_IDENT('africa_adm0'))::regclass,
    (QUOTE_IDENT('amiedes') || '.' || QUOTE_IDENT('air_quality_data_2017'))::regclass,
    (QUOTE_IDENT('amiedes') || '.' || QUOTE_IDENT('air_quality_stations_2017'))::regclass,lass
  );

Now the same query takes ~2 seconds.

[amiedes]

The old sequences_for_table(schema, table_name) did N queries to retrieve all the sequences for all the tables. The new user.db_service.sequences_for_tables(tables) does the same but in a single query.

[amiedes]

The new version of this (sequences_for_tables) has been moved to the DB service.

[amiedes]

These are mostly linter fixes, the important part is under describe '#sequences_for_tables' do

[thedae]

LGTM 👍

[amiedes]

Acceptance in staging ✅

API key creation continues to work as expected:

{
    "id": "75f6c827-6933-4b4f-bf9f-78e29aad4a2e",
    "token": "JWTDtOqbc_d3IogeGIY4wQ",
    "user_id": "8c8ec7bb-29a8-432c-90ee-5d2a1ae472de",
    "type": "regular",
    "name": "Test",
    "db_role": "carto_role_94bb1a0408a21405ba67d88fbb21019b",
    "db_password": "2063f7789c301dc4cc981c560ce0870a992eafdd",
    "grants": [
      {
        "type": "apis",
        "apis": [
          "maps",
          "sql"
        ]
      },
      {
        "type": "database",
        "tables": [
          {
            "name": "dataset_ciudades_2",
            "schema": "amiedes-sequences1",
            "permissions": [
              "select",
              "update",
              "insert",
              "delete"
            ]
          },
          {
            "name": "dataset_ciudades",
            "schema": "amiedes-sequences1",
            "permissions": [
              "select",
              "update",
              "insert",
              "delete"
            ]
          },
          {
            "name": "dataset_ciudades_1",
            "schema": "amiedes-sequences1",
            "permissions": [
              "select",
              "update",
              "insert",
              "delete"
            ]
          }
        ],
        "schemas": [

        ]
      }
    ],
    "created_at": "2020-12-09 09:53:51 UTC",
    "updated_at": "2020-12-09 09:53:51 UTC"
  }

Will do more performance testing in production with the real application.