[mlir][vector] Use dyn_cast instead of cast in patterns

[antiagainst]

Summary

This avoids crashes when the pattern is applied to ops with
tensor semantics.

[mehdi_amini]

Can you provide a test for this?

mlir/lib/Dialect/Vector/VectorTransforms.cpp

 };
 
 // Drop inner most contiguous unit dimensions from transfer_read operand.
 class DropInnerMostUnitDims : public OpRewritePattern<vector::TransferReadOp> {
   using OpRewritePattern<vector::TransferReadOp>::OpRewritePattern;
 
   LogicalResult matchAndRewrite(vector::TransferReadOp readOp,
                                 PatternRewriter &rewriter) const override {
-    auto srcType = readOp.source().getType().cast<MemRefType>();
+    auto srcType = readOp.source().getType().dyn_cast<MemRefType>();
     if (!srcType || !srcType.hasStaticShape())
       return failure();
 
     if (!readOp.permutation_map().isMinorIdentity())
       return failure();
 
     auto targetType = readOp.getVectorType();
     if (targetType.getRank() <= 1)