Restore performance optimizations

[fcostaoliveira]

Using the gist-960-euclidean dataset, and a single docker redis container with 16 threads max configured as follow:

docker run --network=host -d redis:8.0.0
redis-cli config set save ''
redis-cli config set search-min-operation-workers 16
redis-cli  config set search-workers 16

and running:

python3 run.py --engines redis-m-16-ef-64  --datasets gist-960-euclidean  --host 10.3.0.65 --skip-upload --parallels 100 --queries 200000 --no-skip-if-exists --ef-runtime 64

we can see that we can fully saturate Redis by restoring the performance optimizations from PR #16 (85a6bc7) and increase the median QPS from 10K ops/sec to 15K ops/sec.

Summary

Metric	restore-performance-optimizations	update-redisearch	Δ
Median QPS	15,517.66	10,030.81	+54.7%
Median CPU (Peak)	1610.48%	1170.07%	+37.6%