[IMP] crm: improve team pipeline dashboard leads/opps counts performance

The team pipeline dashboard fields
`unassigned_leads_count`, `opportunities_count`, `opportunities_amount`

do a big search on leads, filtered
by type (lead/opp), team, and unassigned user_id (salesman)

Adding an index on user_id, team_id, type significantly
improves the filtering, and therefore the dashboard rendering,
especially when displaying a few teams only
e.g; the teams you are part or responsible of.

The index on `user_id` is removed,
as it become a duplicate as `user_id` is the first column
of this new index.

addons/crm/models/crm_lead.py

@@ -88,7 +88,7 @@ class Lead(models.Model):
     stage_id = fields.Many2one('crm.stage', string='Stage', ondelete='restrict', track_visibility='onchange', index=True,
         domain="['|', ('team_id', '=', False), ('team_id', '=', team_id)]",
         group_expand='_read_group_stage_ids', default=lambda self: self._default_stage_id())
-    user_id = fields.Many2one('res.users', string='Salesperson', index=True, track_visibility='onchange', default=lambda self: self.env.user)
+    user_id = fields.Many2one('res.users', string='Salesperson', track_visibility='onchange', default=lambda self: self.env.user)
     referred = fields.Char('Referred By')
 
     date_open = fields.Datetime('Assignation Date', readonly=True, default=fields.Datetime.now)
@@ -133,6 +133,13 @@ class Lead(models.Model):
         ('check_probability', 'check(probability >= 0 and probability <= 100)', 'The probability of closing the deal should be between 0% and 100%!')
     ]
 
+    @api.model_cr_context
+    def _auto_init(self):
+        res = super(Lead, self)._auto_init()
+        tools.create_index(self._cr, 'crm_lead_user_id_team_id_type_index',
+                           self._table, ['user_id', 'team_id', 'type'])
+        return res
+
     @api.model
     def _read_group_stage_ids(self, stages, domain, order):
         # retrieve team_id from the context and write the domain