[llvm] [CodeGen] Pass alignment and size to supportsUnalignedAtomics. NFC (PR #216621)

[Antonio Frighetto via llvm-commits]

================
@@ -2279,8 +2279,14 @@ class LLVM_ABI TargetLoweringBase {
   /// require a more complex expansion.
   unsigned getMinCmpXchgSizeInBits() const { return MinCmpXchgSizeInBits; }
 
-  /// Whether the target supports unaligned atomic operations.
-  bool supportsUnalignedAtomics() const { return SupportsUnalignedAtomics; }
+  /// Return true if the target supports an atomic access of \p SizeInBytes
+  /// bytes at the given \p Alignment. The default implementation only allows
+  /// naturally aligned atomics, unless setSupportsUnalignedAtomics(true) was
+  /// called.
+  virtual bool supportsAtomicAlignment(Align Alignment,
----------------
antoniofrighetto wrote:

I think either isAtomicAlignmentSupported (or maybe better, supportsAtomicAccess) would read more clearly, but up to you.

https://github.com/llvm/llvm-project/pull/216621