One scoped target for each target definition

By iterating over the uniq pod targets, instead of all pod targets per aggregate target, where now the same PodTarget can occur multiple times.

s/target([ab])/target_\1/

The analyzer generates now first deduplicated pod targets and instantiates them with all target definitions, where they are used, and then generates all the aggregate targets. So it can select the required pod target for each the aggregate target and doesn't need to alter intermediate products as before.

It is computing also the architecture and will compute more.

Xcode insists on these changes.