Here's a paper with a proof that the ground state must be l=0 for spherically symmetric potentials for a single particle, assuming there's a bound state. Abstract:
The variational principle is used to show that the ground-state wave function of a one-body Schrödinger equation with a real potential is real, does not change sign, and is nondegenerate. As a consequence, if the Hamiltonian is invariant under rotations and parity transformations, the ground state must have positive parity and zero angular momentum.
It can be proven that the ground state must be real and non-negative everywhere, assuming a bound state exists and assuming zero spin. The argument in the paper is abstract, but the idea is that, after making the wavefunction into a product of amplitude and phase factors which may vary with position, after calculating the potential and kinetic energy parts of the expectation value $<H>$, the only part of the energy which depends on the phase is a kinetic energy term which is minimized by making the phase constant. If the phase is constant, the wavefunction can be assumed to be real and can't change sign.
From there, it can be proved that the ground state is unique. Basically, if it weren't, then there would be a second ground state wavefunction (which also can't change sign), but they can't be orthogonal since you can't integrate the product of two real non-sign-changing wavefunctions and get zero.
The spherical symmetry of the Hamiltonian demands that if there's an $l \neq 0$ state with an energy, there must also be a $-l$ state with the same energy, so the only possibility is to have $l=0$.
(Having said that, the paper cites this paper, saying that if we have a collection of two particles with strong spin-orbit coupling between them, the overall Hamiltonian can have spherical symmetry, but the system will have spontaneous symmetry breaking and end up with $l \neq 0$.)