Improvements to Elixir benchmarks (#5432)

* Improvements to Elixir benchmarks

1. We increase the pool_size to 40. The original pool_size was
   way too low, which iswas proved by the fact we had to increase
   the queue_target and queue_interval, which means we were often
   waiting more than 5s to get hold of a connection. Also see
   the discussion in #5153. We also rollback the original queue
   values. If we are getting more timeouts, the best solution is
   most likely to bump pool size again.

2. Do not use Repo.checkout. Repo.checkout means we are holding
   the connection for longer, which can be a problem on low pool
   sizes.

* Update page_controller.ex

* Update page_controller.ex

* Update page_controller.ex

* Update page_controller.ex

* Update prod.exs

* Update releases.exs

* Update prod.exs

* Update releases.exs

frameworks/Elixir/phoenix/config/prod.exs

@@ -14,9 +14,8 @@ config :hello, Hello.Repo,
   password: "benchmarkdbpass",
   database: "hello_world",
   hostname: "tfb-database",
-  pool_size: 14,
+  pool_size: 40,
   queue_target: 5000,
-  queue_interval: 5000,
   log: false
 
 config :logger,

frameworks/Elixir/plug/config/releases.exs

@@ -8,7 +8,6 @@ config :framework_benchmarks, FrameworkBenchmarks.Repo,
   database: "hello_world",
   hostname: "tfb-database",
   port: 5432,
-  pool_size: 300,
+  pool_size: 40,
   queue_target: 5000,
-  queue_interval: 5000,
   log: false