events['timestamp'] is nan when data has bmi3d but no ecube data

[katherineperks]

When there is a bmi3d_clock but no sync_clock, ideally events['timestamp'] should be set to events['timestamp_bmi3d']. But in prepare_bmi3d_v1, it continues to get set to events['timestamp_sync']. In entries without sync data, events['timestamp_sync'] is all nan. The tabulate functions use events['timestamp'], so this field shouldn't be left as nan.

[leoscholl]

i think we decided to phase out events['timestamp'] in favor of directly choosing events['timestamp_sync'] or events['timestamp_bmi3d']

maybe we forgot to update some of the tabulate functions?

[leoscholl]

because it's confusing to have 'timestamps' be arbitrarily two different things depending on what data was present