ITIs are not 1s as DEAD_TIME parameter has no effect.

here's the actual dead time inbetween state machine runs:

this code in iblrig.base_choice_world._run() does nothing:

            # handle ITI durations
            if trial_number > 0:
                # The ITI_DELAY_SECS defines the grey screen period within the state machine, where the
                # Bpod TTL is HIGH. The DEAD_TIME param defines the time between last trial and the next
                dead_time = self.task_params.get('DEAD_TIME', 0.5)
                dt = self.task_params.ITI_DELAY_SECS - dead_time - (time.time() - time_last_trial_end)

                # wait to achieve the desired ITI duration
                if dt > 0:
                    log.debug(f'Waiting {dt} s to achieve an ITI duration of {self.task_params.ITI_DELAY_SECS} s')
                    time.sleep(dt)

• Add unit test for time.sleep
• Communicate to users beforehand how this will affect training/animal performance
• Document ITIs by rig version: ITI_duration PP, load_iti function, iblrig changelog, perhaps iblrig docs
• Document parameters in iblrig to make clear how ITI works the difference between dead_time and ITI
• Remove 05-21 and 06-12/001 from 8.28.0 integration data
• Write new test for bpod only habituation cw
• Update ibllib min iblrig version (https://github.com/int-brain-lab/ibllib/blob/78e82df8a51de0be880ee4076d2bb093bbc1d2c1/ibllib/io/extractors/habituation_trials.py#L72)
• QC not informative
• Add warning about ITI < 0.5
• Review imaging task protocols