Even though Rennie and Moretti's answers cover most of the issues connected to this question, I would like to address it more directly, eliminating possible misunderstandings from the beginning.
First, the volume's perspective on mechanics is classical mechanics. No general relativity anticipation was intended.
Second, the ground reference frame, i.e., the reference frame of a laboratory on Earth's surface, is approximately inertial, but not exactly. Deviations from the expected behavior do not depend directly on the presence of gravity but on Earth's motion. If the Earth were not rotating and not revolving around a star in a galaxy, but it was the only planet in the universe, it would behave like an inertial system because a body subject to Earth's gravity plus a compensating force would move with a uniform rectilinear motion. Deviations from the behavior in an inertial reference frame can be detected if the experiments' precision is high enough or the observation time is long enough.
Now, let's go to the Landau- Lifshitz approach. According to the modern view of mechanical phenomena, it is challenging to state Newton's laws satisfactorily.
The main difficulty is in the intertwined role played by the concepts of reference frame, mass, and force. Notice that in Newton's original perspective, the situation was different because the role of reference frames was not so important as after Special Relativity, and also because Newton had his concepts of absolute space and absolute time playing indirectly the role of inertial reference frames.
In Classical Mechanics, inertial reference frames play an important role as the only reference systems where accelerations depend only on real interactions between bodies. A possible way of characterizing an inertial frame is then to say that it coincides with a reference frame far enough from any other body, such that any single particle motion is on a straight line and uniform. The precise formulation may differ, but the essence is this one.
Such a definition may seem to exclude the surface of a planet, where the motion of particles on which only gravity acts is accelerated. However, this is not a real difficulty. Once we have one inertial reference frame where any external force is reduced below the observation threshold by the large distance from other bodies, we know that infinite other inertial frames differ only by the choice of the origin or by the relative uniform velocity. One of them may coincide with the center of a planet not interacting with other bodies.
Landau$Lifshitz's proposal is within the previously described approach. They only shift the emphasis from the characteristics of the motion of a test body to the symmetry conditions allowing them. Notice that from the physical point of view, any statement about space and time, to be amenable to observations must be translatable into a statement about the relative positions of physical systems and physical events. Hidden in their formulation is the asymptotic behavior of a body far enough from any interaction source. Again, once one has found at least one of Landau & Lifshitz's inertial reference frames, there is an infinite number of other inertial reference frames close enough to every possible source of interaction.