Yes. Usually if you use some website (e.g. linkedin), you are logged in. So this does not add any complexity for a user (since he already uses linkedin). At the same time, why would I complicate this and deal with login separately from user workflow. So, i would say it is practical choice. Also, obviously when user is logged-in it helps with anti-bot measures (since user might be asked to click a square or something, so he does that in his usual browser).

Alternative is it do it separately from user flow (in different browser, and solve whole new set of complexities, for no real gain), e.g. in headless browser (which sucks for anti-bot).

I am not really into catch your tail game of anti-bot measures, which are not static but evolve in time. If I tap into what user actually does, I have won a game, no need to re-do it forever.