Request for more information on method to use for setting common config & ENV variables

[lenabanks]

We have a rails application, deployed to Heroku, which we recently integrated DataDog with using the Heroku Build Pack. Our production console logs on 'run' heroku dynos reported multiple intermittent connection errors related to profiling (once per minute at around the same time).
We disable the datadog agent for those type of deployments via the pre_run.sh script (as mentioned in docs here: https://docs.datadoghq.com/agent/basic_agent_usage/heroku/#disabling-the-datadog-agent-for-short-lived-dynos). We assumed the errors were caused by the Agent being disabled successfully but the Profiler and APM logic still attempting to run.

We have not been using the profiling feature so we decided to disable it for that application entirely to reduce the noise on production consoles.
We attempted to disable profiling via setting the "DD_PROFILING_ENABLED" and "DD_APM_ENABLED" environment variables to false by:

1. setting the ENV variable via Heroku
2. setting it programmatically via the pre_run.sh setup script
3. setting via our application config initializer (Rails application)

The third option was the only one that had any effect.

We would love to have more information on which environment variables are able to set appropriately via each of those methods above.
Is there a holistic list and/or documentation that would inform us on which methodologies will / will not work for commonly needed config variables?