Albrecht Durer was a masterful engraver, but the only way he was going to achieve 40 lines per millimetre was by accident. He was working at a time before (just before, but before nonetheless) the Venetian glassmakers' trade secret was leaked and they lost monopoly control of their new cristallo (which was less than 50 years old when St. Eustace was created). That is to say that the lenses of the time wouldn't have been adequate for a loupe one might use to deliberately work at or near the mechanical limits of the plate material. (If Durer had only had the sense to be born 100 years later, the equipment he'd have needed would have been almost commonplace.)
That said, it is not difficult to estimate the frequency of the engraved lines in the work. St. Eustace is approximately 25cm by 35cm, if you count only the image area. (The entire plate is 262 by 361 millimetres.) That's 250mm (horizontal) by 350mm (vertical) to a good enough approximation for the purpose. The detail you posted, then, represents (again, to a "good enough" approximation) about 60mm horizontal and 44mm vertical. And at 1190px by 869px, you can call that 20 pixels (of JPEG) per mm (of physical print), more or less. That tells you immediately, without zooming in and counting pixels, that if you can see line-space-line-space, the absolute maximum that you could discern would be 10 lines per millimetre. Zooming way in, you can see that the most densely-packed areas that are deliberate (that is, they aren't accidents of not-quite-parallel lines, the random mess that is a dead black, or the meeting of adjacent patches of shading) take about 4 pixels to cycle. That's 5 lines per millimetre linearly, or if you're looking at the hatching frequency, 25 (5 vertical and 5 horizontal).
And if that sounds suddenly less impressive, you try it — without lenses or pantographs, on period copper using a period iron or steel (if one could call it that) burin. Millimetres are tiny.