PG: non-blocking server, receive flow control, working tests. I still need to test send flow control.

FIX: bugfix Number.appendHex, this was producing single digits for small numbers. Test for straw man WireParser that would allow to stress test network behaviour

NET: provisional implementation of JsonQueryProcessor - this processor returns query results as json. It is not fully tested yet, but basics are working.

NET: refactored common dispatcher logic into superclass. Fixed couple of places where log was allocated but not sent, this was breaking logging quite significantly

NET: async disconnect, disconnects must be done by the same thread that accepts new connections. Windows dispatcher implementation. Refactored common dispatcher code into super class

NET: async disconnect, disconnects must be done by the same thread that accepts new connections. Linux dispatcher implementation

NET: async disconnect, disconnects must be done by the same thread that accepts new connections. Linux dispatcher implementation

Misc: fixed similar network errors in Windows. Removed duplicate NetFacade. Replaced usages of Os.errno() with interface.

fix: network send() and recv() on Linux do not handle errors correctly, which sometimes results in broken wire protocols.

Misc: fix for get/set lastModified on Windows. Clock change highlighted inconsistent behaviour, which I fixed in this commit.
CUTLASS: fixed TextImportProcessor resume function + test
Misc: Files.exists() does not use Files.getLastModified() anymore

Misc: JMH benchmark for Log4j2 vs QuestDb Log. Good news QuestDb Log is 20x faster and absolutely zero GC
NET: work in progress regarding text file upload, working to test slow/interrupted uploads
Misc: added Os.currentTimeNanos() implementation only for linux for now, will add other OS soon
Misc: commented out legacy test that is failing intermittently. I am rewriting this part of code completely and will focus on new code and tests working consistently