Comparing `hydra()` return gives unexpected results

[woerndl]

I'm experimenting with hydra to show the sections of the current page(s) like so:
Section 2 · Chapter A · 1 — Section 2 · Chapter B · 5

I only want to show the section/chapter part if the content crosses sections/chapters on a given page. Therefore I'm trying to compare if the first use-last: false and last use-last: true elements are the same. Like so:

#context if not (hydra(2, skip-starting: false, use-last: false) == hydra(2, skip-starting: false, use-last: true)) {
  text(red, "This page contains content crossing two chapters")
}

In my understanding on the first page the first and second level 2 headline should be the same element, while on the fifth page content crosses from Chapter B to Chapter C (minimal example not showing my full nesting level):

Full Code Example

#import "@preview/hydra:0.6.1": hydra

#set page(
  width: 120mm,
  height: 120mm,
  header: context {
    if not (hydra(2, skip-starting: false, use-last: false) == hydra(2, skip-starting: false, use-last: true)) {
      text(red, "Page contains content crossing two chapters")
    }
  },
)

= Section 1
== Chapter A
#lorem(200)

= Section 2
#lorem(200)

== Chapter B
#lorem(200)
== Chapter C
#lorem(50)

Shouldn't this comparison also be equal on the first page? Or is there a better way to achieve what I'm trying to do?

[tingerrr]

It seems like a bug in regard to the interaction of skip-starting: false with use-last: true combined with a leading ancestor element (Section 1 for Chapter A), switching them on and off individually reveals nonsensical behavior.

It may be related to #8, but it seems like it's doing the opposite here. Maybe this is strictly on the first page only. I'll have to investigate.