Fix browser profile launch by specifying relative path inside oid dir to profilejob

[tw4l]

Fixes #2911

Follow-up to #2891

This commit strips the oid prefix from the profile filename before passing it along to the profilejob so that it's in the format the crawler expects.

I've also added a profile to an existing test that runs a crawl and verifies that state is complete at finished (i.e. not failed) to catch any similar type regression in the future.

Testing

• Spin up a local instance of Browsertrix
• Verify you can now spin up profile browsers
• Run nightly test that actually launches a profile browser: python -m pytest backend/test_nightly/test_crawl_not_logged_in.py

[ikreymer]

Nice! Waiting for the nightly test to run on this branch to ensure that's also fixed.