Add NumpyGenericOperator.apply_adjoint

[pmli]

See #303

[coveralls]

Coverage decreased (-0.04%) to 81.315% when pulling 3429909 on np-generic-op-apply-adjoint into dfa475f on master.

[sdrave]

LGTM. While we are at it, we might want to hook up a fixture in pymortests.fixtures.operators.

[pmli]

Should there also be source_product and range_product?

[sdrave]

Oh, I totally overlooked that. Yes, since it's part of the interface, it would better to add source_product and range_product parameters.

[pmli]

I have trouble with pickling functions...

[coveralls]

Coverage decreased (-13.9%) to 67.451% when pulling e3ca9cc on np-generic-op-apply-adjoint into dfa475f on master.

[sdrave]

@pmli, we have two tests for pickling: one that allows our hack to make pickling of arbitrary functions possible and one that tests pickling without this hack. The second variant was applied in pymortests.operators to all possible operators. However, lambdas are not picklable, so this test had to fail. I have improved the pickling tests so that everything should be fine now.

[sdrave]

@pmli, feel free to merge ...

[coveralls]

Coverage decreased (-0.1%) to 81.25% when pulling dfc64af on np-generic-op-apply-adjoint into dfa475f on master.