analyze: Group CIDRs with --group

[iBug]

Algorithm:

• Maintain a "grouped keys" bucket
• Start from the sorted keys list (i.e. try to group large items first), for each key:
  • Find its "adjacent part" in the "grouped" bucket
  • If found, try to group the stats recursively
  • If not found, add it unless the bucket is already at capacity, else stop

This algorithm strikes a good balance between accuracy and avoiding too many small items being grouped up into a large one unexpectedly. For example, if there are 256 /24 IPv4 ranges (within the same /16) generating 1 GB traffic and two more /24 ranges generating 100 GB each, with -n 2, the small ranges will not suddenly show up as one /16 with 256 GB traffic and surpass the more problematic ranges with 100 GB traffic.