* Draft of a eigenvalues solver

(does not support complex and does not re-use the QR decomposition)

* Rewrite the cache friendly product to have only one instance per scalar type !
  This significantly speeds up compilation time and reduces executable size.
  The current drawback is that some trivial expressions might be
  evaluated like conjugate or negate.

* Renamed "cache optimal" to "cache friendly"

* Added the ability to directly access matrix data of some expressions via:
  - the stride()/_stride() methods
  - DirectAccessBit flag (replace ReferencableBit)

Eigen/src/Core/Block.h

@@ -71,7 +71,7 @@ struct ei_traits<Block<MatrixType, BlockRows, BlockCols> >
                       || (ColsAtCompileTime != Dynamic && MatrixType::ColsAtCompileTime == Dynamic))
                       ? ~LargeBit
                       : ~(unsigned int)0,
-    Flags = MatrixType::Flags & (DefaultLostFlagMask | VectorizableBit | ReferencableBit) & FlagsMaskLargeBit,
+    Flags = MatrixType::Flags & (DefaultLostFlagMask | VectorizableBit | DirectAccessBit) & FlagsMaskLargeBit,
     CoeffReadCost = MatrixType::CoeffReadCost
   };
 };
@@ -132,6 +132,8 @@ template<typename MatrixType, int BlockRows, int BlockCols> class Block
     int _rows() const { return m_blockRows.value(); }
     int _cols() const { return m_blockCols.value(); }
 
+    int _stride(void) const { return m_matrix.stride(); }
+
     Scalar& _coeffRef(int row, int col)
     {
       return m_matrix.const_cast_derived()