I can answer questions 2 and (tentatively) 4!

When freezing samples, they are cooled rapidly to form vitreous (noncrystalline) ice. If the ice warms enough (and that temp is still well below 0°C), it can transition into a crystalline form. This makes the ice expand and become spiky, which can damage proteins and cells.

For differences in LN2 usage, not every dewar is created equal. Age, the degree of vacuum between the walls, and the distance between the inner and outer walls can substantially affect the thermal conductivity, and thus the boil-off. Differences in how they are capped (which by nature can't be vacuum-insulated) can also change their efficiency.