Allows concatenation to ignore validation and adds method on table to…

[akoumjian]

… separate valid from invalid rows

[moeyensj]

Numpy just makes things so easy sometimes. I was curious to see how you were going to do this with pyarrow compute functions.

[akoumjian]

I implemented it both ways and this appeared the more efficient (both compute and memory) of the two.

[moeyensj]

Nice, there are already tests for these three types:

quivr/test/test_tables.py

Lines 1051 to 1080 in f7c92ae

	def test_apply_mask_numpy():
	values = Pair.from_kwargs(x=[1, 2, 3], y=[4, 5, 6])
	mask = np.array([True, False, True])
	have = values.apply_mask(mask)
	np.testing.assert_array_equal(have.x, [1, 3])
	def test_apply_mask_pylist():
	values = Pair.from_kwargs(x=[1, 2, 3], y=[4, 5, 6])
	mask = [True, False, True]
	have = values.apply_mask(mask)
	np.testing.assert_array_equal(have.x, [1, 3])
	def test_apply_mask_pyarrow():
	values = Pair.from_kwargs(x=[1, 2, 3], y=[4, 5, 6])
	mask = pa.array([True, False, True], pa.bool_())
	have = values.apply_mask(mask)
	np.testing.assert_array_equal(have.x, [1, 3])
	def test_apply_mask_wrong_size():
	values = Pair.from_kwargs(x=[1, 2, 3], y=[4, 5, 6])
	mask = [True, False]
	with pytest.raises(ValueError):
	values.apply_mask(mask)