Step 1: Build the matrix from the rule:
Each entry is the distance between its row number and its column number. That gives 0 on the diagonal, 1 one step away and 2 two steps away. The matrix is $\begin{bmatrix} 0 & 1 & 2 \\ 1 & 0 & 1 \\ 2 & 1 & 0 \end{bmatrix}$.
Step 2: Test symmetry and skew-symmetry by reading rows and columns:
Row 1 is $(0,1,2)$ and column 1 is also $(0,1,2)$. Row 2 equals column 2 and row 3 equals column 3. So $A = A^T$ and $A$ is symmetric. A matrix that is both symmetric and skew-symmetric must be the zero matrix. $A$ is not zero, so $A$ is not skew-symmetric.
Step 3: Test the diagonal shape:
The entry in row 1, column 2 is 1, which is not zero. So $A$ is not diagonal.
Step 4: Test invertibility with row operations:
Row operations of the type "subtract one row from another" keep the determinant the same. First do $R_3 \to R_3 - R_2$, which gives $(1, 1, -1)$. Then do $R_2 \to R_2 - R_1$, which gives $(1, -1, -1)$. The matrix is now $\begin{bmatrix} 0 & 1 & 2 \\ 1 & -1 & -1 \\ 1 & 1 & -1 \end{bmatrix}$. Expand along the first row: \[ 0 - 1(-1+1) + 2(1+1) = 0 - 0 + 4 = 4 \] The value is not zero, so $A$ is non singular.
Final Answer:
Only A (symmetric) and D (non singular) hold, which is option 3.
\[ \boxed{\text{A and D only}} \]