Device recovery tests get stuck on T2B1

[mmilata]

Describe the bug
Some tests in test_recovery_bip39_dryrun.py, test_recovery_slip39_basic.py, test_recovery_slip39_advanced.py get stuck when running on T2B1 hardware. Serial console logs an exception RuntimeError: Waiting for event that already happened - 21 > 1 (different numbers not just 21).

Firmware version and revision
master @ 2195999

To Reproduce
Only seems to happen after several tests have run. E.g. pytest -v ../tests/device_tests/ -k test_recovery_bip39_dryrun is sufficent to trigger this here. Client gets stuck waiting and firmware throws exception:

23327000 trezor.loop ERROR exception:
Traceback (most recent call last):
  File "trezor/loop.py", line 162, in _step
  File "apps/debug/__init__.py", line 127, in return_layout_change
  File "apps/debug/__init__.py", line 117, in get_layout_change_content
RuntimeError: Waiting for event that already happened - 21 > 1