docker-compose container metrics are not organized under docker.container.metric conventions

[ghost]

Issue Description: When tracking metrics from Docker containers created using docker-compose, metrics are deeply nested under the docker.container.[config_hash].etc where config_hash is the unique docker-compose hash for the docker container.

ex:

Sample docker-compose project: docker-compose-test.zip

Steps to reproduce:

1. Download the sample docker-compose project above.
2. Start your containers with "docker-compose up"
3. Start InstrumentalD

Note that metrics collected seem to reflect the inclusion of unnecessary fields introduced by docker-compose environments:

docker.container container_name=instrumental_web_test_1 com.docker.compose.version=1.22.0 com.docker.compose.oneoff=False com.docker.compose.service=web_test_1 engine_host=garden-alpha container_image=api com.docker.compose.config-hash=81bd4d9f263188eff134f6f39a03faf6f22d09396bcb953bae4ebc3acc17e8db com.docker.compose.container-number=1 com.docker.compose.project=api total_pgfault=43854i limit=4143505408i total_pgmafault=11i total_cache=11517952i total_unevictable=0i usage_percent=3.1737903791822446 usage=131506176i fail_count=0i total_mapped_file=1359872i total_rss=119988224i 1538505872000000000

[janxious]

Thanks for the very complete notes. This is indeed not the most useful thing in the Instrumental UI.

For now the workaround will be using the * operator to get metrics to show and we will deal with the bad metric names coming from instrumentald in a future iteration.