Fix panic in AWS ECS detector if container ARN is not valid

[DarthSim]

This PR fixes the panic that may occur here: https://github.com/open-telemetry/opentelemetry-go-contrib/blob/main/detectors/aws/ecs/ecs.go#L116

If containerMetadata.ContainerARN doesn't include : (that is our client case), strings.LastIndex(containerMetadata.ContainerARN, ":") will return -1. This leads to containerMetadata.ContainerARN[:-1], and index out of range panic.

Here's what changes in this PR do:

• Try to extract baseArn from taskMetadata.TaskARN', 'containerMetadata.ContainerARN', and 'taskMetadata.Cluster
• If baseArn is successfully extracted, and some of the ARNs aren't full, fix these ARNs

[codecov]

Codecov Report

Merging #3583 (30decb6) into main (ab49dc8) will decrease coverage by 0.2%.
The diff coverage is 0.0%.

Additional details and impacted files

@@           Coverage Diff           @@
##            main   #3583     +/-   ##
=======================================
- Coverage   80.9%   80.8%   -0.2%     
=======================================
  Files        150     150             
  Lines      10361   10374     +13     
=======================================
  Hits        8388    8388             
- Misses      1831    1844     +13     
  Partials     142     142             

Files	Coverage Δ
detectors/aws/ecs/ecs.go	37.7% <0.0%> (-3.8%)	⬇️

[DarthSim]

Hey everyone!
Sorry for the long delay. I updated the PR:

• Added an entry about the task ARN to the changelog.
• Added test for fixing the task ARN.
• Applied suggested changes.

dismissing my review as changes were applied and I have not time to review it now

[DarthSim]

Hey @pellared @Aneurysm9!

Is there any chance for this PR to be merged? Some of our users face the issue fixed by this PR, so I have to use a fork of this package and update it every time I update the project dependencies; this is pretty annoying.