Improve RDH deserialization
The RDH
load()
(deserialize) implementation was changed to make fewer read calls.
Based on the results of benchmarks (below). X-axis is amount of RDHs deserialized. The from_buf
implementation replaces current
with this merge request.
Benchmarks from running fastpasta on a 2 GB file with the current
implementation and then the from_buf
implementation.