Add configurable response expiration time to Server

[timopohl]

Hey!
I noticed that in our application, when under load, the server sometimes sends and expires the response for a request faster than the client can look up the response.

These are the calls in a working situation:

1594364792.203118 [0 unix:/tmp/redis.sock] "rpush" "routing" "{\"function_call\":{\"name\":\"some_function\",\"args\":[1]},\"response_queue\":\"routing:rpc:C3QZR3XM\",\"timeout_at\":1594364857}"
1594364792.203384 [0 unix:/tmp/redis.sock] "blpop" "routing:rpc:C3QZR3XM" "5"
1594364792.204651 [0 unix:/tmp/redis.sock] "rpush" "routing:rpc:C3QZR3XM" "{\"return_value\":[9999]}"
1594364792.204665 [0 unix:/tmp/redis.sock] "expire" "routing:rpc:C3QZR3XM" "1"

The first two calls are the client pushing a request and immediately block-popping the response, effectively monitoring the list for the response of the server.
But sometimes this situation shows up:

1594364792.179018 [0 unix:/tmp/redis.sock] "rpush" "testing" "{\"function_call\":{\"name\":\"some_function\",\"args\":[1]},\"response_queue\":\"testing:rpc:8K5VTB6O\",\"timeout_at\":1594364857}"
1594364792.185593 [0 unix:/tmp/redis.sock] "rpush" "testing:rpc:8K5VTB6O" "{\"return_value\":[9999]}"
1594364792.185608 [0 unix:/tmp/redis.sock] "expire" "testing:rpc:8K5VTB6O" "1"
1594364793.600321 [0 unix:/tmp/redis.sock] "blpop" "testing:rpc:8K5VTB6O" "5"
1594364798.878143 [0 unix:/tmp/redis.sock] "lrem" "testing" "0" "{\"function_call\":{\"name\":\"some_function\",\"args\":[1]},\"response_queue\":\"testing:rpc:8K5VTB6O\",\"timeout_at\":1594364857}"

So the client first pushes the request, then the server immediately pushes and expires the response before the client calls the "blpop" command on the message queue. At the point where the client calls the "blpop" command, the response is already expired.

The proposed changes offer a configurable expiration duration for the server responses, giving the client more time to get the response back from redis.