Michael Luciuk's answer is right, but there's an even stronger reason for rejecting this hypothesis: refraction in the corona would be wavelength-dependent, but the gravitational bending due to the Sun has been measured over a wide range of wavelengths (at least from visible to radio) and has been found to be independent of wavelength.
Clifford Will's review of tests of general relativityreview of tests of general relativity is a good source for details on this. This particular test is in Section 34.41.1, but the whole thing is worth reading if you want to think seriously about this subject.