add periodic, more descriptive, logging + refactor

• seems the Node heap/memory errors come through before the existing logs that should precede when we're about to exceed the limit; this adds periodic logging in hopes to be alerted before those built-in errors
• additionally, logs will also reveal RSS information and CPU usage when its limit is about to be reached
• although periodically triggered, logs will not emit unless limit is about to be reached (above 80%)

Bug: T378785

Notes: My hunch is that we get those GC-related logs from Node because CPU and/or Heap usage is overburdened and GC is in turn overburdened.

• frequent GC cycles consumes CPU time
• also when heap usage keeps growing and can't get GC'd in time, it leads to out-of-memory issues, which then causes Node to crash bc it can't allocate enough memory