Prior tests

I updated the prior tests to test the specific priors we've implemented. It now tests that the rescale, sample and prob functions work as expected with either float inputs or arrays.

I had to change the prob functions in many cases to allow arbitrary length arguments.

Do we want these functions to work for lists? This is currently not implemented.

@moritz.huebner did I get testing right?

While it doesn't have to be done now. It may be useful to have a test that the probability is normalised. This could then also run on the user input priors.

added 1 commit

• d319303c - add test of normalisation

Compare with previous version

Good idea, I've added a test to do this.

The priors which interpolate from a given input automatically normalise themselves.

It may be possible to use properties to enable changing the limits after instantiation.

mentioned in commit d9cba37f

merged

@colm.talbot Hey just looking briefly through the tests I would just remark that you typically only have one 'assert' statement per test function. That makes it easier to trace down errors later on. Each function should usually contain the minimal code required to get to the 'assert' statement.

That will make the testing code likely much longer and repetitive (which is typically the case for testing code), but that is not an immediate concern.