Revisit MIRAX active position logic
@agpappas noticed that in MIRAX 1.9 and 2.2, the flag byte in the slide position file is reliably 1 if the camera position is active and 0 if not. We can use this to build the active position map in advance, rather than while walking tiles.
On the other hand, in MIRAX 1.8, the flag byte is always 0 and every camera position has coordinates. So, if we want to support MIRAX 1.8, this won't work.
I'm not really aware of the code that does this, but would it be possible to have a separate logic for 1.8? Otherwise, does MRXS 1.8 matter? Is it a popular version?
The separate logic would be roughly what we're doing now, so supporting both approaches would only make the code more complicated. The current code works, it's just somewhat intrusive.
The 1.8 format was already obsolete in early 2010, and probably doesn't matter very much. It also breaks our assumptions in other ways. But we haven't yet ruled out supporting it.