[ty] fix global symbol lookup from eager scopes

[mtshiba]

Summary

cf. #20962

In the following code, foo in the comprehension was not reported as unresolved:

# error: [unresolved-reference] "Name `foo` used when not defined"
foo
foo = [
    # no error!
    # revealed: Divergent
    reveal_type(x) for _ in () for x in [foo]
]

baz = [
    # error: [unresolved-reference] "Name `baz` used when not defined"
    # revealed: Unknown
    reveal_type(x) for _ in () for x in [baz]
]

In fact, this is a more serious bug than it looks: for foo, explicit_global_symbol is called, causing a symbol that should actually be Undefined to be reported as being of type Divergent.

This PR fixes this bug. As a result, the code in mdtest/regression/pr_20962_comprehension_panics.md no longer panics.

Test Plan

corpus\cyclic_symbol_in_comprehension.py is added.
New tests are added in mdtest/comprehensions/basic.md.

[github-actions]

Diagnostic diff on typing conformance tests

No changes detected when running ty on typing conformance tests ✅

[github-actions]

mypy_primer results

Changes were detected when running on open source projects

pandera (https://github.com/pandera-dev/pandera)
- pandera/api/dataframe/container.py:1273:31: warning[possibly-missing-attribute] Attribute `dtype` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1274:33: warning[possibly-missing-attribute] Attribute `parsers` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1275:32: warning[possibly-missing-attribute] Attribute `checks` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1276:34: warning[possibly-missing-attribute] Attribute `nullable` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1277:32: warning[possibly-missing-attribute] Attribute `unique` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1278:32: warning[possibly-missing-attribute] Attribute `coerce` may be missing on object of type `Any | None`
- pandera/api/dataframe/container.py:1279:30: warning[possibly-missing-attribute] Attribute `name` may be missing on object of type `Any | None`
- Found 1636 diagnostics
+ Found 1629 diagnostics

scikit-build-core (https://github.com/scikit-build/scikit-build-core)
- src/scikit_build_core/build/wheel.py:98:20: error[no-matching-overload] No overload of bound method `__init__` matches arguments
- Found 45 diagnostics
+ Found 44 diagnostics

No memory usage changes detected ✅

[mtshiba]

ty_server::e2e pull_diagnostics::document_diagnostic_caching_changed seems to hang non-deterministically.

https://github.com/astral-sh/ruff/actions/runs/19173434027/job/54811898939?pr=21317

[mtshiba]

I found another example of a too-many-cycle-iteration panic.

if C:
    class C[_T](C): ...

The panic occurred because C in the base class list was being resolved circularly.
It should evaluate to Undefined rather than throwing a cyclic-class-definition error.

[mtshiba]

I don't have time right now (I originally found this bug while working on #20566), but I think the logic in infer_place_load is clearly too complicated and could be a breeding ground for bugs like this one.
I think it should be refactored at some point.

[carljm]

The new diagnostic on the conformance suite looks incorrect. We are now throwing an error on the reference to Private at https://github.com/python/typing/blob/main/conformance/tests/generics_syntax_scoping.py#L74 but we should not.

[carljm]

This is looking great, thanks! Just as a heads-up: I am making some (minor: code organization and naming) changes to this PR and then I will merge it.