Make Redis timeout configurable via environment variable

[jbaldassari]

I suspect that we're occasionally seeing Redis timeouts due to messages in our Ceryx logs like the following which are associated with HTTP 500 responses from Ceryx:

[error] 25#25: *41967 lua tcp socket connect timed out
/usr/local/openresty/nginx/lualib/router.lua:5: in function </usr/local/openresty/nginx/lualib/router.lua:1>

The line it's referring to is:

local redisClient = redis:client()

Currently the Redis timeout is hard-coded to 100ms here: https://github.com/sourcelair/ceryx/blob/master/ceryx/nginx/lualib/ceryx/redis.lua#L16

I suspect that we're occasionally seeing new connections hit this limit, so I would like to try increasing it, but unfortunately it isn't currently configurable via Ceryx. This PR adds a new environment variable called CERYX_REDIS_TIMEOUT which can be used to configure this timeout in units of milliseconds. I applied this setting to the Ceryx API as well, although the units had to be converted to seconds in that case because that's what the Python Redis client expects.

[parisk]

Great patch! Thanks for putting the time on this and also updating documentation accordingly.

Merging 🚀!

[jbaldassari]

@parisk and thank you for the quick review and merge!