Added typing spec chapter focused on exception behavior

[erictraut]

Added typing spec chapter focused on exception behavior — in particular, context managers and whether they suppress exceptions.

[hauntsaninja]

I wonder if it's also worth having one sentence explaining why bool | None does not admit the possibility of exception suppression. E.g. in https://discuss.python.org/t/suppression-of-exceptions-within-context-managers-vs-variable-definitions/47256 I think OP was a little confused till linked to the mypy issue where Jelle proposed this heuristic.

[maxfischer2781]

I wonder if it's also worth having one sentence explaining why bool | None does not admit the possibility of exception suppression.

Yes please. Looking at this without knowing its motivation this is extremely confusing otherwise. Starting from the datamodel, one must assume that -> Literal[True, ...] is the important part no matter what the ... is. So -> bool and -> bool | None would behave the same as they both include the case of True.
Explaining that this rule prefers convenience (only the less common suppressing context managers need special handling) over strictness (the more common non-suppressing context managers need precise annotations) would make it clear that this is not an error but by design.

[gvanrossum]

Nits.

[erictraut]

The TC has signed off on this change, so I'm proceeding to merge it.