stats: select ok latency becomes -nan after 30 seconds

[no1seman]

Imagine we have a Cartridge cluster with some data ... if we make the following operations in router console:

unix/:./tmp/run/app-cartridge.router.control> do
                                            >     fiber=require('fiber')
                                            >     crud.cfg({stats = true})
                                            >     t={}
                                            > 
                                            >     res=crud.select('space')
                                            >     table.insert(t, crud.stats().spaces.space.select.ok.latency)
                                            >     fiber.sleep(30);
                                            >     table.insert(t, crud.stats().spaces.space.select.ok.latency)
                                            > end
---
...

unix/:./tmp/run/app-cartridge.router.control> t
---
- - 0.0036880000006931
  - -nan
...

Seems quite strange that "select" "ok" "latency" becomes -nan after 30 seconds