[llvm] [mlir] Shard the public llvm-config.h in multiple files (NFC) (PR #71273)

[Chris B via llvm-commits]

llvm-beanz wrote:

I worry about the tradeoff here. How much does this really speed up builds with a caching layer when you change these settings?

I feel like in a lot of cases, the caching layer should already be resilient against macros that aren't used because the caches tend to identify collisions based on preprocessed files, so if you don't use a macro you should still get a cache hit. This might bypass cache lookups, but it comes at a cost to maintainability.

https://github.com/llvm/llvm-project/pull/71273