PPLNT fix for multiple instances

[crutched]

There is a known issue about "mining time" when you have multiple z-nomp instances (usually for regional stratums) working against one redis. If one miner (with same address) will mine on two or many pool instances all PPLNT system will be broken:

• Wrong maxTime for the round (it will be higher than real time);
• All miners could be affected by < 51% mining time without real violation of PPLNT rules;

Commit also fixes "0" currRoundTime in worker_stats api.

This is important fix for pools with PPLNT turned on and have multiple z-nomp instances working with one redis database in same time.

[appkoders]

Is it safe to apply this update on a running pool? Will my miners lose share or something else?