[mlir] [MLIR][Linalg] Introduce transpose/broadcast semantic to linalg.batch… (PR #130944)

[Andrzej Warzyński]

================
@@ -690,34 +690,32 @@ def MatmulOp : LinalgStructuredBase_Op<"matmul", [
 
     Example Transpose:
     ```mlir
-    linalg.matmul indexing_maps = [
-                   affine_map<(d0, d1, d2) -> (d2, d0)>, // transpose
-                   affine_map<(d0, d1, d2) -> (d2, d1)>,
-                   affine_map<(d0, d1, d2) -> (d0, d1)>
-                   ]
-                   ins(%arg0, %arg1 : memref<5x3xf32>,memref<5x7xf32>)
-                   outs(%arg2: memref<3x7xf32>)
+    linalg.matmul
----------------
banach-space wrote:

> > Does `linalg.batch_reduce_matmul` differ from `linalg.contract`?
> 
> No.

Then why not re-use the definition from `linalg.contract` and begin standarisin things?

> Moreover, I would not prefer to change the documentation of matmul variants in this PR.

I've not noticed anyone proposing this.

https://github.com/llvm/llvm-project/pull/130944