optimize find column index

[Okapist]

Summary

ClickhouseResultSet look for column index on each row. It use expensive equalsIgnoreCase method.
I added index cache for it.

[CLAassistant]

All committers have signed the CLA.

[Okapist]

Collections.synchronizedMap(new HashMap<>()) may be replaced with ConcurrentHashMap if not need to support old jdk versions.

[zhicwu]

Thank you @Okapist for the pull request and sorry for the late reply.

Can we create the HashMap only when needed(e.g. first time calling findColumnIndex)? Besides, when instantiate the cache, can we specify the load factor since we know the maximum number of columns?

Collections.synchronizedMap(new HashMap<>()) may be replaced with ConcurrentHashMap if not need to support old jdk versions.

Perhaps AtomicReference with lazy load will be slightly faster. Having said that, the current implementation is not thread-safe, so I'm good even we use HashMap :)

[Okapist]

Can we create the HashMap only when needed(e.g. first time calling findColumnIndex)? Besides, when instantiate the cache, can we specify the load factor since we know the maximum number of columns?

Good idea. Pull request updated.

[zhicwu]

Thanks @Okapist. Will merge this one after fixing build failure.

[chernser]

Newer PR merged #1568