Potentially. In many areas we have got extremely high test positivity rates, but that is also skewed by the fact that quite a few areas have been switching to symptom-only testing. But all evidence still point to the fact that we undertest more than we did before.
Not necessarily. The issue is that due to increasing vaccinations, the baseline susceptibility of the population shifts. I.e. we do see more vaccinated folks getting infected, but since the vaccine still prevents more serious damage, it might appear milder. What we need to check is basically the outcome among the unvaccinated, but as the number is getting smaller, it is also getting more biased. We have generally more unvaccinated among the younger population, for example, who generally have better outcomes. So we would need to look at unvaccinated folks in an age matched cohort. However, we do not have the data (yet). It would also be important to figure out risk among the elderly, especially if vaccine protection wanes. Moreover, the lung studies were conducted on animal models. So why they provide clues, it is not entirely clear whether it can be translated directly to humans. And especially elderly people are difficult to simulate in animal models.
Together, I would still be somewhat more careful in declaring Omicron less virulent, so far we are looking at milder outcomes, but it may not (solely) because of the properties (i.e. virulence) of the virus, but external factors (vaccination, age) are almost certainly contributing. I am fairly certain that we will get an update relatively soon.
Yes, there is always underreporting. The reported clinical cases are always the minimum known numbers. Depending on testing situation, testing strategies and also willingness to get tested the accuracy of reports has likely varied quite a bit. There is evidence that we have more underreporting than before, though.