Cleanup: Remove duplicate quantization checks (#2566)

The recent upstream
[change](llvm/llvm-project#100667) have
introduced quantization checks that are
already present in the StableHLO core library. This commit removes these
duplicate
checks to avoid redundancy and potential inconsistencies.


|Checks proposed to be removed| StableHLO Code | Upstream MLIR | 
|-|-|-|
| `channel-axis >= 0`|
[cs](https://github.com/openxla/stablehlo/blob/1c0547f391dff5ac71d36dc20a916260afa78c61/stablehlo/dialect/Base.cpp#L795)
|
[cs](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantTypes.cpp#L399)
|
| scale within smallest and largest finite numbers determined by
`expressed_type`|
[cs](https://github.com/openxla/stablehlo/blob/1c0547f391dff5ac71d36dc20a916260afa78c61/stablehlo/dialect/Base.cpp#L765)
|
[cs1](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantTypes.cpp#L327)
[cs2](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantTypes.cpp#L393C9-L393C45)
|


Note that StableHLO has checks like `quantization_dimension <
rank(self)` and
`dim(self, quantization_dimension) = size(scales)` implemented at
[cs](https://github.com/openxla/stablehlo/blob/1c0547f391dff5ac71d36dc20a916260afa78c61/stablehlo/dialect/Base.cpp#L795).
In upstream MLIR similar checks
[cs](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantOps.cpp#L51)
are encoded as part of
[dcast](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantOps.cpp#L110)
and
[qcast](https://github.com/llvm/llvm-project/blob/96f37ae45310885e09195be09d9c05e1c1dff86b/mlir/lib/Dialect/Quant/IR/QuantOps.cpp#L139)
ops and hence cannot be claimed as duplicate.

related upstream clean-up
llvm/llvm-project#110604

stablehlo/dialect/Base.cpp

@@ -751,23 +751,6 @@ bool isValidStablehloQuantizedElementType(Type elementType) {
                   quantizedPerAxisElementType.getScales().end());
   }
 
-  // quantized_type_c6
-  auto maxPosFiniteNum =
-      APFloat::getLargest(
-          cast<FloatType>(quantizedElementType.getExpressedType())
-              .getFloatSemantics())
-          .convertToDouble();
-  auto minPosFiniteNum =
-      APFloat::getSmallest(
-          cast<FloatType>(quantizedElementType.getExpressedType())
-              .getFloatSemantics())
-          .convertToDouble();
-  if (llvm::any_of(scales, [&](double scale) {
-        return scale < minPosFiniteNum || scale > maxPosFiniteNum;
-      })) {
-    return false;
-  }
-
   // quantized_type_c7, quantized_type_c8
   if (llvm::any_of(zeroPoints, [&](int64_t zeroPoint) {
         return storageTypeMin > zeroPoint || zeroPoint > storageTypeMax;
@@ -788,11 +771,11 @@ bool isValidQuantizedDimension(Type type) {
 
   if (!quantizedPerAxisElementType) return true;
 
-  // quantized_type_c11, quantized_type_c12, quantized_type_c13
+  // quantized_type_c12, quantized_type_c13
   int64_t quantDim = quantizedPerAxisElementType.getQuantizedDimension();
   int64_t numScales =
       static_cast<int64_t>(quantizedPerAxisElementType.getScales().size());
-  return quantDim >= 0 && quantDim < rankedType.getRank() &&
+  return quantDim < rankedType.getRank() &&
          (!rankedType.isDynamicDim(quantDim) &&
           numScales == rankedType.getDimSize(quantDim));
 }