For the corrected version, consider the function
$$f(x) = \sup_{\|y\| \le 1} x \cdot y - \frac{1}{2}\|y\|^2.$$
For each given $y$, the function $x \mapsto x \cdot y - \frac{1}{2}\|y\|^2$ is affine, and hence $f$ is convex.
I claim that $f = \mathcal{H}$. Note first that
$$x \cdot y - \frac{1}{2}\|y\|^2 = \frac{1}{2}\|x\|^2 - \frac{1}{2}\|x - y\|^2.$$
Thus, if $\|x\| \le 1$, the maximum of the above expression, over $y$ such that $\|y\| \le 1$, is $\frac{1}{2}\|x\|^2$. Hence $f(x) = \frac{1}{2}\|x\|^2$.
Suppose instead that $\|x\| > 1$. To maximise the above expression, one must minimise the $\|x - y\|^2$ term, again with $y$ restricted to the unit ball, i.e. where $\|y\| \le 1$. This occurs when $y = \frac{x}{\|x\|}$. So, we get
$$f(x) = x \cdot \frac{x}{\|x\|} - \frac{1}{2}\left\| \frac{x}{\|x\|}\right\|^2 = \|x\| - \frac{1}{2},$$
as required. Thus $\mathcal{H}$ is the supremum of affine functions, and hence is convex.