[llvm] [LLVM][AsmParser] Make error reporting of lexer errors more precise (PR #111077)

[Rahul Joshi via llvm-commits]

jurahul wrote:

> Does this help with #29043?

Unfortunately, it does not. The <2 x i32> is a set of valid tokens, so this PR which makes the lexer errors sticky won't address that issue.

https://github.com/llvm/llvm-project/pull/111077