Extension drops platform report metrics under certain conditions

[lahsivjar]

APM AWS Lambda extensiom version (1.1.0 as well as current main):

Describe the bug

• Extension drops the logs if the platform report metrics are reported in an invocation that does not immediately succeed the invocation generating the platform report (relevant code).
• Extension always drops the last platform.report metric for an execution environment.
  - [ ] Extension drops the platform.report metric for the last invocation if this invocation times out/crashes. #360 (the logs API doesn't report this at all; further discussion can continue on the dedicated issue)

To Reproduce
Steps to reproduce the behavior:

1. Run a lambda function along with the extension.
2. Invoke the function a bunch of times.
3. Observe cloudwatch logs for a warning message report event request id didn't match the previous event id.
4. Observe the number of metrics with faas.billed_duration : * to be equal to number of invocations - number of coldstarts.

Expected behavior
Extension doesn't drop platform report metrics and sends it to APM-Server.

Provide logs (if relevant):

^ Logs from cloudwatch showing dropped platform reports (the logs is observed using a lambda extension from a dev branch)

[AlexanderWert]

@lahsivjar Also in case that a function is only called once per lifetime (so every request is a cold start) lambda metrics get lost completely. With this issue we should also make sure that cached data is flushed when the Lambda runtime is going to be shut down.