Table.copy() changes sparsity

`Table.copy` can sometimes change sparsity, which is unexpected. It is likely needlessly complex as it goes through the constructor (`Table.__new__`). I found this in one of our tests:

```
import unittest
from Orange.data import Table
from scipy import sparse as sp

from Orange.widgets.tests.base import WidgetTest
from Orange.widgets.data.owmergedata import OWMergeData

class TestOWMergeDataBug(WidgetTest):

def test_sparse(self):
self.widget = self.create_widget(OWMergeData)
data = Table("iris")[::25]
data_ed_sparse = Table("titanic")[::300].to_sparse()
self.send_signal("Data", data)
self.send_signal("Extra Data", data_ed_sparse)

output_sparse = self.get_output("Data")
self.assertTrue(sp.issparse(output_sparse.X))

output_sparse = output_sparse.copy()
self.assertTrue(sp.issparse(output_sparse.X)) # FAILS

if __name__ == "__main__":
unittest.main()
```

Start with the Table.copy and Table.__new__ entry points named in the issue, then run the provided TestOWMergeDataBug.test_sparse reproduction. Trace why copying output_sparse changes the sparsity of X, and verify that the sparse assertion passes after the change.

Written by the indexing model from the issue text.