Reduce allocation pressure in DBScan refill by reusing buffers

[asmit27rai]

Fix: #195

How to test it:

1. Build

make clean && make -j4

2. Start server

./strfry relay (Terminal 1)

3. Benchmark

python3 scripts/populate_test_db.py --relay ws://localhost:7777 --events 10000
python3 scripts/write_rate_sweep.py --relay ws://localhost:7777 \
       --rates 100,500,1000 --duration 30 --output result1.json

4. Full Benchmark

python3 scripts/populate_test_db.py --relay ws://localhost:7777 --events 20000
python3 scripts/write_rate_sweep.py --relay ws://localhost:7777 \
       --rates 0,100,500,1000,2000,5000 --duration 60 --output result2.json

5. Correctness Check

python3 scripts/populate_test_db.py --relay ws://localhost:7777 --events 5000
python3 scripts/capture_req_results.py --relay ws://localhost:7777 \
       --filter '{"kinds":[1],"limit":100}' \
       --filter '{"kinds":[0,3],"limit":50}'

[asmit27rai]

@hoytech As discussed, here is a approach i tried with some tests.
Please test it locally.
Thank You.