[EXPERIMENT] chore: run e2e with chromedriver

[sadym-chromium]

Experiment. https://github.com/GoogleChromeLabs/chromium-bidi/actions/runs/9303439623

Performance:

1. Ubuntu old headless chromedriver works ~3:30 - 4:00 min per shard vs 2:50 - 3:15 in mapper.
2. Ubuntu new headless chromedriver works ~1:15 - 1:30 min per shard vs 1:00 - 1:15 in mapper.
3. Mac os headful chromedriver works ~0:55 - 1:10 min per shard vs :52 - 1:10 in mapper.
4. Ubuntu headful chromedriver didn't work at all (timeed out in 20 minutes)

Failing tests:

• Headful ubuntu hangs forever:

session not created: Chrome failed to start: exited normally.
(session not created: DevToolsActivePort file doesn't exist)
(The process started from chrome location /home/runner/.cache/chromium-bidi/chrome/linux-127.0.6508.0/chrome-linux64/chrome is no longer running, so ChromeDriver is assuming that Chrome has crashed.)

• tests/browsing_context/test_close.py::test_browsingContext_close
• tests/browsing_context/test_close.py::test_browsingContext_close_prompt
• tests/script/test_get_realms.py::test_scriptGetRealms
• tests/session/test_protocol_basics.py::test_binary
• tests/session/test_protocol_basics.py::test_channel_empty
• tests/session/test_protocol_basics.py::test_channel_non_empty
• tests/session/test_protocol_basics.py::test_empty_object
• tests/session/test_protocol_basics.py::test_invalid_json
• tests/session/test_protocol_basics.py::test_session_status
• tests/session/test_subscription.py::test_subscribeToMultipleChannels_eventsReceivedInProperOrder
• tests/browsing_context/test_get_tree.py:test_browsingContext_afterNavigation_getTreeWithNestedCrossOriginContexts_contextsReturned

Issues:

• Static commands do not support BiDi+ channels
• Empty BiDi+ channel should be considered as missing channel
• Navigating to a page with OOPiF via BiDi command hangs