[jdbc-driver] Synchronize access to the handlers map

[guilgaly]

This does seem to solve the intermittent test failures in https://github.com/zengularity/query-monad as we discussed.

[cchantep]

Thx. What's the test failing w/o? To possibly add a unit test there?

[guilgaly]

I'll see about adding a unit test, shouldn't be too hard.

In query-monad, the issue can occasionally happen in any one of the tests in WithPlayTransactionSpec. The handler registration is executed on the current thread, and the JDBC query execution happens on a different thread which may or may not yet see the previous mutation of the handlers map.

[guilgaly]

I added a simple test which basically just reproduces the same situation we have with the problematic query-monad test, except with many iterations. When executed a few times on my computer, it did consistently fail with the non-synchronized HashMap and succeed with the synchronizedMap.