[OPT-1591] Update import benchmark description in the documentation

AT:

the import benchmark contains only two data sets: One that is currently called large and will be named medium dataset. A new data set with 10 million process instances, which is the the large data set
for both dataset I can see a distribution of process defintions and some statistics, e.g. number of process instances, decision instances, activity instances, variable instances, process definitions, decision definitions
I can read about how long it took to import this data and what was the average amount of data imported per minute or second
I can read about how much resources I need for Elasticsearch, Optimize (RAM, Cores) to get those results

Camunda Optimize