We don't for absolute certainty, but there's no reason to think there would be. Consider:

a) A virus that is known to cause severe disease in a significant proportion of people, including months-long serious health effects, hospitalisation and death.

b) A virus chosen because it doesn't causing any ill-effects in humans, modified to take on some non-harmful aspects of virus a) and found to not cause any significant harm in tens of thousands of people over several months.

Which is more likely to cause long-term side effects?

I appreciate the need for caution, but these people are experts in what they do, any doubts we might have would have already been considered by them.

We can't. If everybody turns into superheros in 10 years, we wouldnt know; or maybe this is actually the backstory of 'children of men'; we can't know. RNA vaccines are researched in animals for about a decade, but long term effects are usually not beeing found in such studies (disection usually cuts that short). There is though no theoretical nor experiental evidence for that.

Most adverse events (narcolepsy and GBS included) occur within a few months (1-3) of administration. The fact that some are reported after years doesn't mean they are "long term": it only means it took a while to vaccinate those people. So far none of the trials (even more so AZ) has incurred in vaccine-related adverse events of this sort.

If there is no known mechanism by which side effects could occur with long gestation, and if experience with other vaccines and vaccine trials provides no evidence of any such, then it would seem unlikely.