Max heap size does not match the actual maximum allocated memory

[fulmicoton]

(thank you for the amazing project!)

I am running bytehound on a multithreaded rust server. There are no child process.
While running, I see resident memory using top reaching 8GB. This is not off-heap data: jemalloc's allocated memory (exposed via an endpoint on my server) also reaches close to 8GB.

However, bytehound reports suggests a peak of around 1GB.
Where could the discrepancy come from?

Does bytehound drops allocation if it cannot keep up with the logging?
Is there a way to do sampling over the allocations?