OK - first, the H-alpha issue:
DSLRs incorporate an Infra-red blocking filter, so that the colours come out correctly for normal shots, without being affected by infrared wavelengths.
Unfortunately, the normal filters used for this in DSLRs also block about 80% or so of the deep red hydrogen alpha light.
Most of the astrophoto images you see with a lot of h-alpha red in them have been taken either with modified DSLRs (where the IR blocking filter has been replaced with one that passes h-alpha) or dedicated monochrome astro cameras used with filters (a different filter for each colour channel - either R G B (for normal colour images) or narrowband (typically h-alpha, oIII and sII) filters for emission nebulae, with an optional extra luminance image). There are also a couple of DSLR models intended for astrophotographers that have h-alpha friendly filters already
installed in place of the usual ones. (Downside of these and the modified DSLRs is that it throws the colour balance off for normal photography, though you can largely compensate by using a custom colour balance - or fully compensate with the equivalent of the original filter).
So, with an unmodified DSLR that's blocking about 80% of the H-alpha, you're not going to see anywhere near as much h-alpha red as you will from a modified or mono camera.
Now onto stacking:
You seem to be misunderstanding what stacking does.
It's not about increasing the effective exposure - it's all about reducing noise.
An astrophoto image consists of the wanted signal, plus unwanted noise.
Since the noise is generally random + or -, if you stack (basically average) a lot of
images, the random noise gets reduced (by a factor depending on the square root of the number of images combined - so 4 images will halve the noise, 16 will quarter it, 64 will reduce it to an eighth and so on.
(There are also time dependant errors - that's what "dark" and "bias" frames are used to compensate for).
Note that what you get out of the stacking process is NOT a brighter image - it's a less noisy one. What that means is that you can then stretch the (brightness range of the) image more in post processing before the remaining noise becomes a problem.
So for your first question, NO - if you're using the same exposure time for the sub images, the ISO 3200 subimages will be twice as bright as the ISO 1600 ones (as you'd expect). When you stack them, the brightness of the result will still be about the same as the sub images - so the ISO3200 one will be twice as bright as the ISO 1600 one - but the noise level will have gone down more compared to the originals in the ISO 1600 result, since you're stacking more images (9 would give you 1/3 the noise, compared to 1/2 the noise with 4).