Consider making jemalloc metrics expensive

[rkuris]

Some concern about how expensive it is to gather jemalloc metrics.

If this doesn't cause any noticeable slowdown after merging #1837 then we can consider closing this.

Original comment follows:

That is only true for Gatherer but this breaks the moment someone start's using metrics server

firewood/ffi/src/metrics.rs

Line 82 in b11fc0e

pub fn setup_metrics_with_exporter(metrics_port: u16) -> Result<(), Box<dyn Error>> {

I think we should add this under experimental or expensive flag (as you noted below), I cannot think of a scenario where we'd be collecting jemalloc stats from the fleet of nodes in production, it would mostly be noise.

I think this is good segway to start separating experimental and production metrics.
Not going to block on this but, it would be good to open a follow-up issue that would add this behind expensive metrics or maybe even introduce new flag.

Originally posted by @Elvis339 in #1837 (comment)