ggr-ui keeps polling selenoid nodes which are removed from quota xml files

[vgrigoruk]

1. Remove 1 server address from quota.xml
2. Reload ggr-ui config via docker kill -s HUP ggr-ui-container-id
3. Load http://your-ggr-ui:8888/status

• Expected: instant response with /status response aggregated from currently running selenoid instances.
• Actual: ggr-ui responds after some time (few seconds), as it is trying to poll /status endpoint for non-existing selenoid instance.

GGR-UI logs:

2018/05/23 13:16:29 [STATUS] [Failed to fetch status from http://10.0.185.224:4444: Get http://10.0.185.224:4444/status: dial tcp 10.0.185.224:4444: i/o timeout] [10.0.160.134]

Relates to #5

[vgrigoruk]

it seems like logic for loading/merging quota xml should be different for ggr-ui, comparing to ggr (in case when servers are deleted from quota)