[DOC] Enhance Memory Footprint Document

[singhmanas1]

Add the memory footprint for the following workload:

1. Streaming cagra index build -- build ivf-pq in batches and run cagra::optimize() [Relevant to Serverless deployment]
2. Disk based cagra index build - building knn graph with data streaming from disk.
3. CAGRA --> HNSW conversion

Document: https://docs.rapids.ai/api/cuvs/stable/neighbors/cagra/#memory-footprint