cheating? you must be kidding! You should never be rigid in your thinking as a programmer.
it's best to be an outside the box thinker!!!

Why was the kata still on Node 8?

What is happening here?

I would like to know the usecase of this "feature"

I have no idea why it works

O(1) solution

it need not be a consideration. Node 8 is only supported for legacy reasons. if you use padStart(), the runner will take notice that the kata no longer compiles with Node 8 and disable that version.

(besides, Node 8 does support padStart())

This kata allows Node v8.0 which doesn't support padStart

The numbers are still not zero-padded correctly.

What you're looking for is ...

    let input = Array.from({length: random(2,5)}, () => {
      const l = random(1,1e6).toString();
      const r = random(1,1e6).toString();
      const len = Math.max(l.length, r.length);
      return l.padStart(len,'0') + ' ' + r.padStart(len,'0');
    }).join("\n");

Better yet, have a numbers with variable length in the random tests (ported from Python / Ruby translations)

    let input = Array.from({length: random(2,5)}, () => {
      const n = random(1, 10);
      const l = random(10**(n-1), 10**n-1).toString();
      const r = random(0, 10**n-1).toString().padStart(n,'0');
      return random(0, 1)==1 ? l+" "+r : r+" "+l;
    }).join("\n");

padStart is available natively. Why reinvent the wheel?

I've created a fork with a fix

(The user attempted in JavaScript.)

In the assumptions:

• both numbers in the pair will be of the same length
• If a number is shorter, it will be zero-padded.

The actual input:

Input:
537302 92264

Something is wrong I guess

Same with js