fixed

This is not the case. For example, x=2.
The action lambda x:0 results in 0,
and the action lambda x:x%2 results in 0 too

There is a trouble with python random test case -
a test should look as a tuple:

(command, number_to_operate_with, expected_result_of_operation, "explanation");

the command is in range 0..4. But in the test case we see random.randint(0,1000000)

It should be:
random.randint(0,1000000) % 5
or:
random.randint(0,4)

Nobody's going to change the requirements now. Apart from that, such inputs would contradict the idea of the kata.

Fixed no.1

Thank you ))

Added to description.

it means "don't hard code everything" (I guess...)

There are two things.
First, the description is misleading because of a typo:

'Punctuation marks should be treated as if they are apart of the word'

should read:

'Punctuation marks should be treated as if they are a  part of the word'

not apart which may be understood as separately, but a part of word, which means together with word

The other thing is about tests. The tests should include strings with leading / trailing spaces and multiple spaces between words.

You write:
TestError: Expected: '', instead got: ' ' - expected an empty string with no characters, even a single space, and got a string of one character, space char.
It's a special case and should be dealed separately.