Earth's gravitational field causes Earth to retain a gaseous atmosphere, which both absorbs light itself and refracts light towards the surface. Estimating the altitude of the optically thick part of the atmosphere as somewhere between 6 km and 60 km, this atmosphere effectively increases the cross-sectional area of the Earth for interacting with sunlight by between 0.1% and 1%; the lower end is a better estimate. Not all of this atmospherically captured sunlight is absorbed by the Earth, but the same is true for directly incident light as well.
So, interaction with Earth's gravitationally-bound atmosphere increases insolation by something like 0.1%, subject to local, daily, and seasonal fluctuations due to things like clouds.
Atmospheric refraction of sunlight in the ideal case bends the light by about half a degree, or 1800 seconds of arc. In the same ideal configuration, the general-relativistic deflection of light by the sun is 1.75 seconds of arc. Scaling the GR deflection by $M/R$ for mass $M$ and radius $R$ to about 0.6 milliarcseconds, I get that the atmospheric refraction is about three million times larger than the general-relativistic refraction.
So general relativistic refraction of light is a parts-per-billion corrections to Earth's insolation. Not important.