IO gas cost estimation based on DB accesses

[jakmeier]

Building on top of #7053, instrument our estimations such that they count the DB accesses. Based on that, give a report based on fixed-delay assumptions for DB accesses.

This kind of estimation has already been done by manually extracting the information from the traces. The tricky part is to remove the warm-up iterations, the accesses only done for setup, and to figure out by which number to divide the total number of accesses. The idea behind this issue is to automate this process.