Please refer to the linked documentation why rounding numbers is a bad idea for a kata.

You are not studying a questionnaire, you are authoring a kata. These things are not the same.

Your creativity does not trump established quality guidelines for authoring a kata.

If you want whole numbers, go study a questionnaire.

But I want whole numbers.
What do you mean "no point"?
This is a creative process.
When studying questionnaires, most of them give whole numbers with rounding.
This is exactly what I wanted the result of the kata to look like.

There is no reason to ask for integers. You can ask for fractions or ( approximate ) floating point values. In the case of floating point approximations, handle assertions according to the documentation that was even linked for you above.

It's too late for this kata, but the concept is applicable to all kata that use floating point values. Note that this is a complication that is probably best avoided for your next first kata.

Updated the description in accordance with your wishes.
As I understand, the question was related specifically to the description, in this case.

How do you think the specification should be transformed to make it clear?
The condition states that the percentage is presented as a whole number without rounding.

The response is expected to be a list of integers containing the percentage of votes for each item, without rounding.

The percentage may not be an integer.

ok i changed description/condition, solution and test stories. removed rounding.

Did you read that documentation, and understand that the problem is that fractions cannot be exactly represented in floating point numbers?

In short, no. Fractions may be a possibility though. Or comparing results with an error margin.

I cannot see your reply.

I must therefore assume the issue has not been resolved and reraise it. I'd rather not do that dance. Does your reply really need to be spoilered?

If we use the "decimal" module, would it be the right solution for this version of the kata in your opinion?

This documenatation: https://docs.codewars.com/authoring/recipes/floating-point#do-not-use-rounding-or-stringification-to-work-around-problems-with-precision

What documentation?
The method is built into the standard Python interpreter package.

An unlimited number of choices also means an answer is not possible.

This requirement cannot be correct.

If the number of survey items is to be unlimited, separate arguments are not a good choice. An iterable would be an appropriate one.