Workaround slow compile times for big StaticArray

[j8r]

Fixes #2485.

I didn't add details on the commit message, having no knowledge of what is causing this slow compile times in the background with LLVM - be free to edit it if somebody wants.

[waj]

Please, don't merge this yet (#2485 (comment))

[waj]

This should be now superseded by #9510 (if specs passes 🤞 😄 )

[j8r]

Closing in favor of #9510, which also reduces the compile time.

[j8r]

I think it can be fine to have this simple fix for now, before a true codegen fix (eventually by @waj ?) will be ready.
There is no downsides, and the changes are minimal.

[waj]

I'm sorry @j8r but I still don't think it's a good idea. This is basically always unrolling the contents of the block. It might look harmless for small examples but the block could be any piece of code. Whatever the block is, the codegen will be writing LLVM code for it N times.

[j8r]

Ok, but block are always inlined, so is there really a difference at the end?

[waj]

Block calls are inlined but the code from the block is only generated as many times yield is called. LLVM might decide later to unroll loops, effectively repeating the code, but that probably depends on the number of iterations and the size of the block.