unskip test_admin_edits_hotp_secret and test_admin_edits_totp_secret

[redshiftzero]

Description

As part of #4974 we had to temporarily skip two tests (test_admin_edits_hotp_secret and test_admin_edits_totp_secret) that failed CI when we ran using the latest version of Tor Browser which were hanging CI indefinitely. We should investigate why these tests are hanging and resolve. See also #4691 where similar but intermittent behavior was also reported.

[garrettr]

I'm stepping through test_admin_edits_totp_secret in pdb and it's hanging in _admin_visits_reset_2fa_totp on self.retry_2fa_pop_ups(_admin_visits_reset_2fa_totp_step, "reset-two-factor-totp").

[garrettr]

Just logging some progress here in case someone else wants to pick up where I've left off.

In _admin_visits_reset_2fa_totp, clicking the "reset 2fa" button via self.safe_click_by_id("button-reset-two-factor-totp") is expected to trigger the JS confirmation dialog registered in journalist.js.

I walked through test_admin_edits_totp_secret in pdb and found that clicking button-reset-two-factor-totp takes ~1-2 seconds to return. Once the pdb prompt returns, p self.driver.page_source shows that the form was submitted and the next view, admin.new_user_two_factor, was rendered. It seems the expected confirmation dialog was not shown, so the test gets stuck in retry_2fa_pop_ups.

Next, I watched the test over VNC. I didn't see the confirmation dialog there either, although VNC was a little slow/glitchy, so it's possible I missed it.

This behavior suggests a possible cause: if journalist.js is not loading or executing in the test Tor Browser, then what I saw is what you would expect: clicking the form submit button would go directly to the next page in the 2fa reset flow, without being held up by a confirmation dialog that requires JS to exist.

I tried repeating the steps of the test by running make dev and using Tor Browser 9 on my Mac to walk through a 2fa reset flow, and got the expected JS confirmation dialog, so I think this issue is unique to the functional test environment (or possibly Linux?).

Somewhat desperately, I tried updating to the latest geckodriver (v0.26.0), but that didn't resolve the issue.

I think the next best thing would be to get some more useful logs (e.g. web console, network) out of the browser and Selenium webdriver. Does anybody know how to do that?

[garrettr]

The skipped tests pass if I switch the default driver for the functional tests from Tor Browser to Firefox ESR, which suggests this issue is specific to Tor Browser or tbselenium.

[zenmonkeykstop]

This makes sense, there was another test that was failing (rather than hanging) because Javascript was being disabled in TBB. PR #4987 will fix that one, and if it passes muster the same fix will work here.

[zenmonkeykstop]

Grabbing now that #4987 is merged.