Call lver_to_energy_bound with the signature it has

[Chessing234]

lver_to_energy_bound(hypotheses, tau0) builds the tau0 <= tau <= 2 tau0 and 2 <= tau <= tau0 domains itself, but three call sites still pass the older (LV*, LV_zeta*, sigma_interval) triple and raise ValueError: Parameter hypotheses must be of type Hypothesis_Set. on the first line they reach. So derived.prove_heath_brown_energy_estimate, derived.prove_improved_heath_brown_energy_estimate and examples.zero_density_energy_examples cannot run at all, and the first two are the \derived \code{} links for two theorems in the zero density energy chapter.

Ported to the current API, dropping the domain/compute_LV_star blocks the callee now does itself, and printing the bounds as the prove_zero_density_energy_* functions do.

prove_improved_heath_brown_energy_estimate needed a judgement call: it built two hypothesis sets, one for LV* and one for LV_zeta*, which the current API cannot express. I merged them (k = 2..4, Huxley, Heath-Brown region 2a, the zeta large value estimate, both lver conversions). It reproduces the theorem the blueprint attaches to it, crossover point included:

A*(x) <= (18 - 19x)/(2(3x - 1)(1 - x))   on [3/4, sqrt(13889)/328 + 143/328)
A*(x) <= 4(10 - 9x)/(5(4x - 1)(1 - x))   on [sqrt(13889)/328 + 143/328, 4/5)

which is Theorem imp-hb-energy-bound. That function also passed sigma in [1/2, 1] while its tau0 was only defined on [3/4, 4/5]; the interval now comes from tau0.

prove_heath_brown_energy_estimate likewise reproduces hb-energy-bound:

A*(x) <= (10 - 11x)/((2 - x)(1 - x))     on [1/2, 2/3)
A*(x) <= (18 - 19x)/(2(2 - x)(1 - x))    on [2/3, 3/4)

and then the two improved branches above, ending at 12/(4x - 1) on [5/6, 25/28) — matching the theorem's third branch.