we should check item bounds for Projection where-clause well-formedness

[lcnr]

see the related zulip thread for more info https://rust-lang.zulipchat.com/#narrow/channel/144729-t-types/topic/want.20help.3A.20check.20item.20bounds.20for.20Projection.20where-clause.20WF/with/554688567

fixes the following issues:

• No error reported when a generic parameter doesn't meet the requirement of an associated type #78893
• ICE: delay_span_bug: error performing ParamEnvAnd #103899 (fixed via a workaround in Emit dropck normalization errors in borrowck #136539)
• ICE in mir_drops_elaborated_and_const_checked #135039 (fixed via a workaround in Emit dropck normalization errors in borrowck #136539)
• 1.92 regression in PyO3: unable to determine layout for ... because ... cannot be normalized #149081
• Delayed bug on uhh nested associated types with implied traits? #149120

[lcnr]

cc @rust-lang/types

[Mattias-Petersson]

Hi. I am entirely new to open-source and contributing. I think this could be a good issue to make into my first contribution :D I will look into this first on the weekend but it sounds interesting.

@rustbot claim

[Mattias-Petersson]

Wait this was marked as hard, not easy. I apologize :( I will find a more suitable first task!

@rustbot release-assignment