go/types, types2: audit type assertions for potential Alias bugs

[findleyr]

#67540 fixes a bug where GODEBUG=gotypesalias=1 (the default in go 1.23) results in a type checker bug.

While reviewing the fix (https://go.dev/cl/587075), I noticed that right above that bug, there is another, allowing the following program to compile without error (the error comes from vet, not type checking, which may actually indicate another bug: is vet not honoring GODEBUG?):

https://go.dev/play/p/cy0Jma_6jAA?v=gotip

It appears that we need to more thoroughly audit type assertions within the type checker.

CC @griesemer @adonovan

[griesemer]

There are many assertions with type parameters that seem incorrect. I will take this on.

[gopherbot]

Change https://go.dev/cl/587156 mentions this issue: go/types, types2: add missing Unalias calls (clarification)

[gopherbot]

Change https://go.dev/cl/587157 mentions this issue: go/types, types2: don't panic converting a constant to aliased type parameter

[gopherbot]

Change https://go.dev/cl/587176 mentions this issue: go/types, types2: coreType/String must consider Alias types

[gopherbot]

Change https://go.dev/cl/587158 mentions this issue: go/types, types2: underIs must consider Alias types

[gopherbot]

Change https://go.dev/cl/587159 mentions this issue: go/types, types2: operand.convertibleTo must consider alias types

[gopherbot]

Change https://go.dev/cl/587161 mentions this issue: go/types, types2: operand.AssignableTo must consider Alias types

[griesemer]

With these most recent CLs I believe the major issues here are covered.
There are still a few cases (in type inference, in the mono checker) which need attention, but I am not yet sure they are wrong. I will try to create test cases for these as well.

But I think for the freeze we're ok on this front. We may find more regressions but once found, they are easy to address.