CI: Checks job aborted if a step fails

[datapythonista]

In our job that checks that the code is ok (flake8, mypy...) [1], we want to run all the steps whether any of them fail or not.

This is implemented by using if: true in each step. While this should work, it doesn't. A better way to implement this that probably fixes the issue is to use if: always() instead.

What we should do is:

• Replace all instances of if: true by if: always in the yaml file.
• Introduce couple of temporary errors. Something PEP-8 invalid (so flake8 reports the error), and change the order of the imports of a file (so isort reports the error).
• Open a pull request
• See if the job behaves as expected, and wait for a code review
• Once the pull request is approved, revert the errors introduced, so the changes to the yaml file can be merged

1. https://github.com/pandas-dev/pandas/blob/master/.github/workflows/ci.yml

[suzutomato]

I'd like to take this.

[datapythonista]

Thanks @suzutomato

You can assign the issue to yourself by writing take in a comment.

[suzutomato]

Sorry, I misunderstood the instruction in the contributing guide.
It had to be exactly "take", not including the exact word "take" in my comment.

[suzutomato]

take