Random tests are comparing the user's results against itself:

    expected = space_age(rsa, rp)                          # user's result
    test.assert_equals(space_age(rsa, rp), expected)       # compared against the user's result...

See this solution or this one...

There're already a few very similar katas available. In fact, I wouldn't call it any different from Multiply...

The values should be compared for approximate equality.

Duplicate: https://www.codewars.com/kata/56b22765e1007b79f2000079

odd length arrays in JS :
eg [ 110, 252, 250, 255, 117 ] Test Passed: Value == '[110 | 252], [250 | 255], [117]'

There's a preloaded list of your chief’s seven favorite string delimiters called: delimiters

What's the point of this in the notes? Isn't it passed to the function as an argument directly?

The user can modify the input in Python.

The user can modify the input in JavaScript.

When witing things intended to be strings in the Description, please format them as strings: "[1 vs. 2], [3 vs. 4]" ( just as you are doing for the function argument ).

I don't like the way the inner delimiter is specified either but the outer delimiter ( which is always ", " ) is left implied. We're supposed to reverse-engineer that from a couple of examples, and that's akin to asking us to read your mind.

There are odd lengths in JS.

PHP translation added, please check.

well, the kata is retired, but FYI:

• would have need random tests
• the variable name is still camelCased in the solution setup
• not enough tests, by far! Amidst those, theses edge cases are lacking:
  • empty string?
  • string beginning with char of codepoint below 64 or between 91 and 96 (included) => that would require some additional information in the description: are we dealing with variable names or any kind of strings? in the first case, the only possible edge case is "_" as first char.

There's a lot of katas about counting items matching a predicate; nothing new here.