[ty] Improve diagnostic range for non-subscriptable diagnostics

[AlexWaygood]

Summary

Currently our diagnostic only covers the range of the thing being subscripted:

But it should probably cover the whole subscript expression (arguably the more "incorrect" bit is the ["foo"] part of this expression, not the x part of this expression!)

Test Plan

Added a snapshot

Co-authored-by: Brent Westbrook 36778786+ntBre@users.noreply.github.com

[astral-sh-bot]

Diagnostic diff on typing conformance tests

No changes detected when running ty on typing conformance tests ✅

[astral-sh-bot]

mypy_primer results

Changes were detected when running on open source projects

pandas (https://github.com/pandas-dev/pandas)
- pandas/tests/series/test_arithmetic.py:646:17: error[non-subscriptable] Cannot subscript object of type `bool` with no `__getitem__` method
+ pandas/tests/series/test_arithmetic.py:646:16: error[non-subscriptable] Cannot subscript object of type `bool` with no `__getitem__` method
- pandas/tests/series/test_arithmetic.py:647:21: error[non-subscriptable] Cannot subscript object of type `bool` with no `__getitem__` method
+ pandas/tests/series/test_arithmetic.py:647:20: error[non-subscriptable] Cannot subscript object of type `bool` with no `__getitem__` method

sympy (https://github.com/sympy/sympy)
- sympy/parsing/autolev/test-examples/ruletest6.py:31:6: error[non-subscriptable] Cannot subscript object of type `<module 'math'>` with no `__getitem__` method
+ sympy/parsing/autolev/test-examples/ruletest6.py:31:5: error[non-subscriptable] Cannot subscript object of type `<module 'math'>` with no `__getitem__` method

No memory usage changes detected ✅

[carljm]

I think highlighting the whole expression (as you do here) is best. It is both the type of the base object (which is the type that is not subscriptable), and the fact that you are subscripting it, that's "at fault" for the error. (In the case of not-subscriptable, the actual key is arguably not involved at all! But it still makes sense to highlight the entire expression IMO.)

[MichaReiser]

Similar to attribute expressions, I think it's undesired to highlight the entire object because it can get very noisy if that expression is long (think a call expression)

I think the ideal is to highlight .attribute or [value].

Related discussion astral-sh/ty#169

I don't think this should block this PR because it at least makes it consistent but we should update said issue to also mention subscript

[AlexWaygood]

Heh, well, I disagree for the same reasons I gave in astral-sh/ty#169 ;) but I'll add a note in that issue to mention that the same discussion also applies to subscript diagnostics!