Probably unused ring buffers for Prometheus Config

[nowkarol]

Describe the bug
Hi, I've got question (not sure if it is a bug) regarding ring buffers in AbstractTimeWindowHistogram and TimeWindowMax. I use them with Prometheus so my DistributionStatisticConfig consists of 1 minute expiry and buffer length of 3. I have some doubts about using 3 histograms.

As far as I can tell values are recorded in each one of them. Reads are done only from currently selected. Every 20 seconds during histogram interaction previous one is cleared same as accumulated histogram. I cannot see any usage of other two histograms.

So my question is: how (for Prometheus) are used two other histograms? Is this oversight and PrometheusMeterRegistry should override bufferLength to 1?

Additional context
This task and this one are concerned with TimeWindowMax. @shakuzen answer is correct but don't point why it behaves like this, it is also little confusing regarding that comment - it doesn't mention buffer size dependency.

[nowkarol]

Sorry for spam, took me a while to understand that it's done to group (and evict) values in smaller intervals than expiry time