Yes I think it is advisable to split your tests in small(ish) modules, each testing a certain aspect of your code. That way you have a greater chance to "recycle" these tests for other code: e.g. a test to check if the module loads; another to test whether object are properly instantiated; if standard input is giving the expected results; if edge cases are properly handled; ... You will probably need a similar (but not exactly the same) suite of tests for many of your scripts. Having a score of "standard" tests available (a "test library") would help you install tests easier and with more confidence.

Putting this all in one large test script is asking for trouble: after a while you would need a test to check the tests!.