Speed up list by using compacted strings

[JackyWoo]

Which issues of this PR fixes:

This PR try to fix #285

Change log:

Beanchmark:

only list request 
150 threads

1. List result contains 100 items, each item size is a random value range from 10 - 100 bytes.

List RPS increases from 458433 to 619388 and the TP99 looks better.

before

read requests 14826483, write requests 0, Read RPS: 458433, Read MiB/s: 2441.74
Read sampler:
0%		0.18 msec.	
10%		0.212 msec.	
20%		0.22 msec.	
30%		0.228 msec.	
40%		0.235 msec.	
50%		0.244 msec.	
60%		0.254 msec.	
70%		0.269 msec.	
80%		0.296 msec.	
90%		0.596 msec.	
95%		1.016 msec.	
99%		1.515 msec.	
99.9%		2.262 msec.	
99.99%		2.973 msec.	
{"timestamp":1718169452404,"read_results":{"total_requests":14826483,"requests_per_second":458432.9663788887,"bytes_per_second":2560348117.2260935,"percentiles":[{"0.00":0.18},{"10.00":0.212},{"20.00":0.22},{"30.00":0.228},{"40.00":0.235},{"50.00":0.244},{"60.00":0.254},{"70.00":0.269},{"80.00":0.296},{"90.00":0.596},{"95.00":1.016},{"99.00":1.515},{"99.90":2.262},{"99.99":2.973}]}}

after

read requests 14172371, write requests 0, Read RPS: 619388, Read MiB/s: 3156.67
Read sampler:
0%		0.177 msec.	
10%		0.207 msec.	
20%		0.215 msec.	
30%		0.221 msec.	
40%		0.226 msec.	
50%		0.233 msec.	
60%		0.239 msec.	
70%		0.247 msec.	
80%		0.257 msec.	
90%		0.274 msec.	
95%		0.293 msec.	
99%		0.381 msec.	
99.9%		1.268 msec.	
99.99%		5.157 msec.	
{"timestamp":1718165728991,"read_results":{"total_requests":14172371,"requests_per_second":619387.5307082109,"bytes_per_second":3310006964.104679,"percentiles":[{"0.00":0.177},{"10.00":0.207},{"20.00":0.215},{"30.00":0.221},{"40.00":0.226},{"50.00":0.233},{"60.00":0.239},{"70.00":0.247},{"80.00":0.257},{"90.00":0.274},{"95.00":0.293},{"99.00":0.381},{"99.90":1.268},{"99.99":5.157}]}}

2. List result contains 10 items, each item size is a random value range from 10 - 100 bytes.

RPS decreases from 750492 to 728299

before

read requests 14011512, write requests 0, Read RPS: 750492, Read MiB/s: 449.475
Read sampler:
0%		0.153 msec.	
10%		0.175 msec.	
20%		0.181 msec.	
30%		0.186 msec.	
40%		0.191 msec.	
50%		0.195 msec.	
60%		0.199 msec.	
70%		0.205 msec.	
80%		0.212 msec.	
90%		0.224 msec.	
95%		0.236 msec.	
99%		0.283 msec.	
99.9%		1.005 msec.	
99.99%		1.711 msec.	
{"timestamp":1718169245436,"read_results":{"total_requests":14011512,"requests_per_second":750491.8734469928,"bytes_per_second":471308896.52471157,"percentiles":[{"0.00":0.153},{"10.00":0.175},{"20.00":0.181},{"30.00":0.186},{"40.00":0.191},{"50.00":0.195},{"60.00":0.199},{"70.00":0.205},{"80.00":0.212},{"90.00":0.224},{"95.00":0.236},{"99.00":0.283},{"99.90":1.005},{"99.99":1.711}]}}

after

read requests 15859158, write requests 0, Read RPS: 728299, Read MiB/s: 527.171
Read sampler:
0%		0.151 msec.	
10%		0.177 msec.	
20%		0.183 msec.	
30%		0.188 msec.	
40%		0.193 msec.	
50%		0.198 msec.	
60%		0.203 msec.	
70%		0.209 msec.	
80%		0.218 msec.	
90%		0.232 msec.	
95%		0.248 msec.	
99%		0.383 msec.	
99.9%		1.099 msec.	
99.99%		2.594 msec.	
{"timestamp":1718169076273,"read_results":{"total_requests":15859158,"requests_per_second":728299.1706786874,"bytes_per_second":552779070.5451238,"percentiles":[{"0.00":0.151},{"10.00":0.177},{"20.00":0.183},{"30.00":0.188},{"40.00":0.193},{"50.00":0.198},{"60.00":0.203},{"70.00":0.209},{"80.00":0.218},{"90.00":0.232},{"95.00":0.248},{"99.00":0.383},{"99.90":1.099},{"99.99":2.594}]}}

In summary when list response contains more items the optimization looks better.

[JackyWoo]

The following is the flame graph when benchmarking with 100 items. We can see after optimization list request processing takes less CPU cycles.

1. before optimization

2. after optimization

[lzydmxy]

LGTM