$\text{Rep}(D(G))$ as representation category of a vertex operator algebra
This answer is related to my answer here: Duality between orbifold and quasi-Hopf algebra (twisted quantum doubles) and the comment by Scott.
Every finite group $G$ can be embedded in some symmetric group $S_n$ and then you can take for example the $n$-fold product of the $E_8$ lattice VOA, for which the representation category is trivial. Then I guess the $G$ orbifold of this VOA should have $\mathrm{Rep}(D^\omega(G))$ as a representation category. This would be true in the setting of conformal net, see my answer here: Duality between orbifold and quasi-Hopf algebra (twisted quantum doubles) but I am not sure how much a result like: "an extension of a rational VOA is given by a special symmetric Frobenius algebra $A$ in the category of Representations and the Representation category of the extension is given by the the subcategory of dyslexic (local) module category of $A$." is established in VOA.
In this case the results mentioned in http://arxiv.org/abs/0909.2537v1 also show that every VOA with $\mathrm{Rep}(D^\omega(G))$ as representation category is a $G$ orbifold of a holomorphic VOA.
Update I could still not find an exact statement in VOA, but in Alexei Davydov wrote in http://arxiv.org/abs/1312.7466 (p.2.)
Our examples come from permutation orbifolds of holomorphic conformal field theories (CFTs whose state space is an irreducible module over the chiral algebras) . It is argued in [30] (see also [35]) that the modular category of the $G$-orbifold of a holomorphic conformal field theory is the so called Drinfeld (or monoidal ) centre $\mathcal Z(G,\alpha)$, where $\alpha$ is a 3-cocycle of the group $G$. It is also known that the cocycle $\alpha$ is trivial for permutation orbifolds (orbifolds where the group $G$ is a subgroup of the symmetric group permuting copies in a tensor power of a holomorphic theory). he assumption crucial for the arguments of [30] is the e xistence of twisted sectors. This assumption is known to be true for permutation orbifolds [1].
This statement (if true, I did not yet check [1] (http://link.springer.com/article/10.1007/s002200200633)) gives exactly the answer the ops question, like I guessed above, namely:
Let $G\subset S^n$ and and let $(V^{\otimes n})^G$ the $G$-permutation orbifold of $V^n$ where $V$ is a holomorphic VOA, then $\mathrm{Rep}((V^{\otimes n})^G) \cong \mathrm{Rep}(D(G))$.
Update: The statement "an extension of a rational VOA is given by a commutative algebra $A$ (Thm 3.2) in the category of Representations and the Representation category of the extension is given by the the subcategory of dyslexic (local) module category of $A$. (Thm 3.4)" appeared now in http://arxiv.org/abs/1406.3420
A lot has happened in the last four years, and we now have lots of positive results.
The current state of knowledge is given in Evans-Gannon, "Reconstruction and Local Extensions for Twisted Group Doubles, and Permutation Orbifolds". In particular, if $G$ is a finite solvable group, then $D(G)$ (and more generally, any twist $D^\omega(G)$) is the representation category of some vertex operator algebra (in particular, the fixed points of a $G$-action on some holomorphic vertex operator algebra).
For non-solvable groups, the result you want would follow from the conjectured regularity of fixed points (i.e., a suitable generalization of C-Miyamoto).
Oddly enough, it turns out that permutation orbifolds can have non-trivial twists. This is discussed in the Evans-Gannon paper, and earlier in Johnson-Freyd's "The Moonshine Anomaly".