[CIR][Transforms] Simplify redundant bitcasts

[seven-mile]

Fix #479 .

There are three available stages to place the simplification in.

• A straightforward method is to extend fold method for CastOp.
  But CIR does not use CanonicalizerPass, so it does not work.
• As for somehow equivalent to it, append a pattern to MergeCleanupsPass.
  But now it is mainly for CFG-related simplifications like block merging. I don't know if this is the proper way. Shall we rename it to a broader definition?
• Add a new pass for this issue.
  This is definitely not very reasonable XD. We won't consider it unless we're really out of options.

This PR includes the second option. What do you think @bcardosolopes ?

[seven-mile]

Oh, it seems applyOpPatternsAndFold will perform DCE on the cast ops that we want to simplify. This broke the CIR/CodeGen/unary.c test. I think the testcase is to blame for unused variables, right?

[bcardosolopes]

Thanks for your first PR, way to go. See comments inline.

[victor-eds]

What would be the most appropriate spot for a standalone test for this? Something like:

// CHECK-LABEL: @ptrbitcastfold
//       CHECK:  %[[ARG0:.+]]: !cir.ptr<!i32>
//       CHECK:   return  %[[ARG0]]
func.func @ptrbitcastfold(%arg0: !cir.ptr<!i32>) -> !cir.ptr<!i32> {
  %0 = cir.cast(bitcast, %arg0) : !cir.ptr<!i32>, !cir.ptr<!i32>
  return %0 : !cir.ptr<!i32>
}

[seven-mile]

Thank you all for reviewing and providing so many constructive suggestions! Apart from a few typos, I've tried to address the two NFC commits for testcases. I'm not very familiar with llvm-lit, so please point out any issues if you notice them.

[seven-mile]

I just realized that I probably shouldn't have placed the standalone test under CIR/CodeGen. I'll first merge it into clang/test/CIR/Transforms/merge-cleanups.cir.

[bcardosolopes]

Awesome, thanks!