If a reviewer opens a pull-request in browser, then comments or approves it, and meanwhile the pull-request is updated, the comment or approve is applied to the new pull-request state without reviewer noticing it.

Steps:
1. Developer creates a pull-request
2. Reviewer opens the pull-request in browser
3. developer push another commit C to the pull-request.
4. reviewer comments a change in the pull-request, which is changed in the commits C

result: the comment fails to bind to a code
expected: it should be shown in a context with note "Outdated"

the approve variation, I think it is serious:
4a. reviewer approves the pull-request

result: pull-request is approved without reviewing C
expected: reviewer must look at the updated PR to be able to approve it. PS: In this case, auto-unapproving plugin was installed, it did not help.