Provide cgroup aware machine metrics

Currently FDB collects some machine metrics like memory which work fine for normal setups. Once you start to run FDB inside containers (or `cgroups`) you will get wrong/confusing machine metrics since some of the metrics are not `cgroup` aware like `/proc/meminfo` https://github.com/apple/foundationdb/blob/master/flow/Platform.actor.cpp#L454 this will provide the host machine metrics and not the cgroup metrics. An example output of FDB running in Kubernetes (with a memory limit of 8GiB):

```bash
Process performance details:
1.1.1.1:4500:tls ( 2% cpu; 1% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 98.9 GB RAM )
1.1.1.1:4502:tls ( 1% cpu; 1% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 98.9 GB RAM )
1.1.1.2:4500:tls ( 1% cpu; 9% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 133.3 GB RAM )
1.1.1.2:4502:tls ( 2% cpu; 9% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 133.3 GB RAM )
1.1.1.3:4500:tls ( 2% cpu; 11% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 92.4 GB RAM )
1.1.1.3:4502:tls ( 1% cpu; 11% machine; 0.001 Gbps; 0% disk IO; 0.4 GB / 92.4 GB RAM )
```

One approach would be to check if `/sys/fs/cgroup/memory/memory.stat` exists and if so use the metrics contained there (we would need to adjust the parsing).

Start in flow/Platform.actor.cpp around line 454 and inspect how machine memory metrics are read from /proc/meminfo. Compare that path with the cgroup memory metrics under /sys/fs/cgroup/memory/memory.stat, including the required parsing changes. Done means containerized FDB reports cgroup-aware machine metrics instead of host memory values.

Written by the indexing model from the issue text.