Method: Solve the leaf/internal-node condition as an inequality directly in terms of n, without substituting a specific number.
A node at index i in a 1-indexed, n-element array heap is a leaf if and only if its left child index 2i exceeds n: \( 2i > n \), i.e. \( i > n/2 \).
Since n is odd, n/2 is not an integer; it lies exactly between the consecutive integers \( \frac{n-1}{2} \) and \( \frac{n+1}{2} \). So the condition \( i > n/2 \) for an integer index becomes \( i \geq \frac{n+1}{2} \).
Equivalently, index i is an INTERNAL node exactly when \( i \leq \frac{n-1}{2} \).
Compare each option's index against the boundary \( \frac{n-1}{2} \):
(A) \( \frac{n+1}{2} = \frac{n-1}{2}+1 \), strictly greater than the boundary: LEAF index, not selected.
(B) \( \frac{n-1}{2} \), exactly equal to the internal-node boundary: INTERNAL index, selected.
(C) \( \frac{n-3}{2} = \frac{n-1}{2}-1 \), strictly less than the boundary, comfortably internal: selected.
(D) \( n \), far greater than \( \frac{n+1}{2} \) for any n > 1, and always the last node of the heap with no possible children: not selected.
So the indices that fail to be leaves are exactly \( \frac{n-1}{2} \) and \( \frac{n-3}{2} \), options (B) and (C).
\[ \boxed{\text{Options (B) and (C)}} \]