Dots on repeats do not line up with barlines when using Mensurstrich

What's the expected behavior ?

Centered to the barline, I's say

From my experience (and as the original implementor of the Mensurstrich feature), I don't think there is an "expected behaviour". Mensurstrich is a purely typographic device used only in modern editions of early music; its purpose is to add hints for measure bars when the original source was lacking them (as it was customary until approx. first half of XVII c.), without breaking too much the "free flow" of "un-barred" music printing.

@mike320 : repeats WERE used "back then", under several forms, occasionally needing some interpretation (and raising very scholarly arguments ;) ); for instance attached is a potential source of the example in the OP (from Dowland's First Booke of Songes originally published in 1597, here from a 1613 reprint), which shows the barred Cantus and the unbarred Tenor, both with a final repeat indication.

Being present in the source, the repeats should not be rendered with Mensurstrich, but with regular on-staff measure bars, which would avoid any problem with the dots. From my experience, I would assume this to be by far the most common case.

Sure, one case remains uncovered: when a repeat is an editorial addition, but I am far from sure that repeat dots between the staves would look any better than those shown in the OP; even the repeat bars do not look very nice between the staves and contradict the Mensurstrich purpose, which is to have the measure bars as incospicuous as possible. Personally I would engrave the (editorially added) repeat normally within the staves, adding about it a note in the apparatus.

In summary, it seems to me that:

1) it is quite rare a case;
2) the proposed solution is not significantly better than the current one

and this does not seems to justify the added complexity of managing it.

@mike320 : sure, you are right, Mensurstriche are a modern compromise, to add bar lines -- with the minimum visual impact -- where there was none, at the same time indicating that they are added.

One can argue that this does not really apply to repeat signs, but only to bar lines. If the repeat signs are original, they are not "added" and should not go between staves, but within the staves (together with their (double) bar lines), where they originally were.

If they are an editorial addition, then I have no obvious solution, but I would like to see a real case, before spending time and increasing the code complexity without a compelling reason (the example proposed by the OP seems an especially contrived case; certainly, nothing in the source of Come again suggests that repeats should be added there).

While mike320 conclusion is probably reasonable, I am still convinced that there is no "real" bug: while I did not foresee this specific development while implementing the feature (as I never expected someone would use it in this non-standard way), Mensurstrich is not actually intended to be applied and should not be applied to repeats; it is intended as a philological notation and the feature is currently as it is essentially by design.

In other words, the repeats should remain within the staves, regardless of the position of other bar lines. Does the OP knows that each bar line can have its own span independently of the other, isn't it?
If an arranger feels to add repeat indications, he is obviously free to do it, but an arrangement is hardly the context where to use philological notations...

Generalising and as an aside: what about arbitrary alterations to the span of a repeat? If you alter it to encompass only part of a staff, would you expect the dots to move up or down to remain centred on the line segment? To me it would look very odd.