Bread

An autolyse does a couple of things: it starts gluten development and the breakdown of the flour starches into simple sugars (which provides food for the yeast). It means you won't have to work the dough as much when you come to knead, which helps reduce oxidation. These processes rely on reasonably high hydration of the flour, so if you're doing a poolish there usually isn't enough flour and water left over to autolyse. With their high hydration, poolishes and preferments will provide some of the benefits of an autolyse anyhow (though the yeast fermentation inhibits some of the chemistry); if you really wanted you could do an autolyse of the poolish before adding the yeast. I've never bothered, but now I'm intrigued.

More autolyse info here: https://www.bakerybits.co.uk/resources/autolyse-what-why-how/

Bit more complicated. Salt is hygroscopic and has an effect on the absorption of water into the flour, which in turn will affect things like gluten development (it's complicated, and I can't confess to understand all the mechanisms). To make things even more confusing, salt actually improves the dough structure overall.

The reason it's not so important to autolyse with the poolish is because there's not much more water to be absorbed by the flour, and it is the effect of salt on absorption that we're concerned about.

Say you have a recipe like so
Poolish: 500g water , 500g flour
Final dough:
500g flour, 200g water, 1kg poolish.

Because there's only 200g remaining of water, you'd only be benefitting from 200g of water autolysing in 500g flour, while you've essentially already had 500g of each flour and water getting an extended autolyse in the poolish

But say for example you made a recipe with a smaller proportion of poolish for whatever reason (slightly self defeating as generally a poolish should use half of the total flour to get the distinctive flavour and properties).
EG
Poolish: 150g flour, 150g water
Final Dough: 850g flour, 550g water, 300g poolish.

You'd probably want to autolyse, as there's a lot of flour and water left to incorporate.

If you're going to bake the same recipe again, you could always see if it benefits from an autolyse, but I'd guess it would be negligible.

Edit, TBH I think @Thrustvector has explained it better.

Edit 2: it was also noted that for slowly fermented breads, with reasonable hydrations >70%~, that the effect of autolyse was negligible on final product (according to experiments done for Modernist Baking, IIRC). I'm not sure if it factored in differences for mixing by hand though, was more looking at production baking.