dnd 3.5e – How exactly do cylinder spells and effects “ignore obstructions” in their areas? Or what kinds of obstructions they can realistically ignore?

I think this is a poor description, because it is trying to describe a method of drawing a particular volume (by drawing a horizontal two-dimensional shape on the $XY$ plane and projecting it downward on the $Z$ axis), but then also using that as the description of what the spell itself is doing. The reason that’s a problem is because the combination of two factors here undercut any concept of the spell being “projected” at all. The first issue is that spells come into effect instantaneously; there is no time at which an effect is just the circle and then it travels downward. The other issue is the last quoted line, “A cylinder-shaped spell ignores any obstructions within its area,” which says we have to ignore anything blocking the projection on its way downward.

The sum of those two effects is that the spell isn’t “projected” at all, but simply comes into existence at all points within the volume simultaneously, without regard for what else is in that volume or where any point within is in relation to any other point in it. This is a valid way to imagine magic happening, of course, and probably appropriate for many spells. Moreover, even when it isn’t precisely appropriate, the top-down projection is very specific and definitely isn’t going to apply in a lot of cases, so it makes sense to avoid implementing that as a global default. All of which they could have easily just said, and relied on the “projection” description to talk solely about how you should draw the shape/know you have the right shape, but they didn’t do that. Instead, they talk about the “spell” doing that, and then need to apply the “hack” of having it ignore obstructions.

Anyway, yes, the long and short of it is, the official rule is that, by default, cylinder-shaped spells simply fill the entire volume without regard for what’s in it, or where any point is in relation to another or the existence of anything between them. Which is kind of nonsensical for some spells, but appropriate for others, so we should look to individual spell descriptions creating exceptions for those spells that actually want effects to come down from above. (Of course, I imagine, a great many authors of individual spells didn’t double-check the default rules and think through these concerns, and probably many spells that should have such an exception lack it—that’s where the DM comes in.)