Based on reading some documentation on Metalink, the answers to your questions are
1. RMAN doesn't require a large pool allocation, unless you have i/o slave processes enabled by question 2. Of course you may still need a large pool allocation for other reasons.
2. The i/o slave processes that are relevant to RMAN are backup_disk_io_slaves and backup_tape_io_slaves. One reason for using these parameters is if you are running on an O/S that either doesn't support asynchronous i/o, or implements it inefficiently. The goal would be to improve throughput of your backup jobs.