add runtime metrics

[rochdev]

This PR adds runtime metrics that are captured every 10 seconds. The feature is behind an experimental flag until it will have been finalized.

The following groups of metrics are captured:

• CPU utilization
• Memory usage
• Heap statistics
• Heap space statistics
• Event loop statistics
• Garbage collector statistics
• Process uptime

A runtime ID was added to both the traces and the metrics so that we can correlate them in a later version.

For the moment, the following limitations exist:

• Prebuilt binaries are not published, so the C++ add-on is built on the machine running npm install.
• Compilation is only tested on the official Docker Node images.
• Network statistics are not available.
• File system statistics are not available.

[willgittoes-dd]

Looks good to me :)

[pawelchcki]

One small suggestion to stop the callback once EventLoop is deleted. Otherwise looks great!