Fix slow response

[longhoangwkm]

Problem:

curl -XGET localhost:8432/a/1AuGvfDfNhbeVNUDzXHjLmEnFnWv3pkjjG/utxos

Response time: ~60 seconds or time out.

Debug:
To get utxos, we searchRange all blocks by addressToScriptID:
If latest block is ~1M6. Split all blocks to n time to hit db by chunkSize.
Example, latest block is ~1M6. / chunkSize(10) = 160k times.
i found that chunkSize too small, hit to leveldb too much.
Reduce n to help reduce response time

Solution:
I tried to increase chunkSize value to 100, 1000, 10_000 and 100_000.
Benchmark show that with chunkSize = 10_000 return good response Time
< 1 seconds on mainnet

$ time curl localhost:8432/a/1AuGvfDfNhbeVNUDzXHjLmEnFnWv3pkjjG/utxos
real	0m0.638s
user	0m0.005s
sys	0m0.015s

$ time curl localhost:8432/a/1Afb76aMKPTpSppV9nZrf3Nfmg4KXvikeA/utxos
real	0m0.169s
user	0m0.006s
sys	0m0.005s

$ time curl localhost:8432/a/19s5TAWJNdsRxzoHN51TPsWnCieD2h1M1y/utxos
real	0m0.090s
user	0m0.006s
sys	0m0.004s

[jdogresorg]

I just noticed this pull requests and applied the suggested fix to the indexd instances on my servers and it seems to be running MUCH faster. Thank you very much @longhoangwkm for submitting this fix.

@chiguireitor ^^

[longhoangwkm]

You're welcome. Btw, we sometimes have problems with exceeding the max limit for the number of node object errors. I think we should keep developing this CIP: https://counterpartytalk.org/t/cip22-change-indexd-to-addrindexrs/5870. We need a big change, I'm eager to contribute to a new project.