Purpose
=======

Calling copy in a list comprehension or a loop does break the batch definition of many stuffs (create, message_subscribe, followers management, ...) leading to a large amount of unnecessary requests while duplicating many records.

The copy method is actually the only method left from the old API that does not handle recordsets.

This commit is making the copy and copy_data methods work in batch improving the execution of the computed fields, and the whole records creation process.

Now, copy works mainly like this:

@api.returns('self')
def copy(self, default=None):
    vals_list = self.copy_data(default)  # this one returns a list of vals
    return self.create(vals_list)

And if you want specific vals for each record in the output, simply use copy_data() and modify its output accordingly, like in:

@api.returns('self')
def copy_data(self, default):
    vals_list = super().copy_data(default)
    return [dict(vals, name=_("%s (copy)", product.name)) for product, vals in zip(self, vals_list)]

For the record, duplicating 400 project.tasks (with lot of postprocess on copy about milestones, task dependencies, SOL mapping) divides the number of sql request by 2.4 (11000 -> 4500) and divides the execution time by 3 (15 -> 5 seconds).

TaskID: 3742289