A Counterexample to Horseshoe Analysis of Indicatives

The horseshoe analysis identifies indicatives as having the same semantic content with material conditionals. But is that right? Look at the following conditional:

C: "If I ate one and only one egg this morning, I ate two eggs this morning."

C is obviously false, actually necessarily false. But suppose I did not eat any egg this morning. Then, given the horseshoe analysis, C is true because it has a false antecedent and false consequent. So the horseshoe analysis is committed to treating something necessarily false as true. This I take to be evidence against the horseshoe analysis.

Of course, a horseshoe analysis may retort to my example this way. Since I know that I didn't eat any egg this morning and indicative conditionals do not tolerate known-to-be false antecedents, C is inappropriate. But this reply may commit the horseshoe analysts to much bigger problems, I reckon.