Totally agreed.
Although, a stereo compressor can still reduce the stereo image by turning the differences in the left and right sides the same amount. By doing this, especially at high GR, you will lose the differences between the left and right dB by dB. So yes, it may not reduce the stereo image in a certain sense (as in the way you explained), but by the nature of compression, you will lose dynamics which can only lead to losing dynamics that contribute to the stereo image.
Its a hard concept to verbalize, but the more you play around with compressors, the more you notice things like these. (and I'm still far from the level of understanding of some of the people on this board). But to put it simply, if the left and right are compressed the same amount, thats like putting a mono reverb on a stereo track; essentially making the left and right more of the same.
Both methods are capable of totally professional results, but as usual in the audio production world, its completely subjective and up to the person behind the board to make the best decision to serve the bigger picture. So try out both and make the best choice :Smokedev: