While I agree that what you suggest partially solves the problem I outline, I have several issues with that approach:

First, if used as a .ansible-lint-ignore, as opposed to a .config/ansible-lint-ignore.txt, it only works on filenames in the current directory, but not path names. e.g.:

playbook.yml name[play]

works ... but this does not work:

path/to/playbook.yml name[play]

That means you need to have directory specific versions, unlike say .gitignore. Plausible workaround - use symlinks to a single copy, I guess.

Second, when the Ansible Lint docs say "ignore", the lint rules defined like those don't really "ignore" that check. Instead, it is marked as a WARNING, not an ERROR. The upshot of that is that the ignore-d rule is still in the output, and then downstream tools, such as VSCode, still pick up on it and present it as a problem. That's not really "ignored". This is, as you mentioned, already discussed here:

#3068

Third, this feature require each file to be enumerated.

IMO, it would be way nicer if ansible-lint followed the yamllint config idea:

https://yamllint.readthedocs.io/en/latest/configuration.html#ignoring-paths