AntiAliasing: restore var binding in mapApplication

[mario-bucev]

PR #1507 made some changes to mapApplication, and one I made by negligence (in the diff line) turned out to unveil an issue in the Let case handling of AntiAliasing (line in question).

Here, we compute the targets of the transformed binding. If this binding happens to contain a function call, the returned targets may be invalid because getTargets works on functions prior to the AntiAliasing function purification, not after (as it's done here).

The reason this issue did not arise before is the var introduced by mapApplication prevented target computation, which ultimately resulted in rejection.

This PR simply reverts the changed lines but the underlying issue is still here (though it cannot be triggered). To properly fix this, we would need to distinguish pre/post transformation getTargets computation, which would require significant changes to EffectsAnalyzer.

[vkuncak]

Could you move the rest of the PR comment to the /* ... */ in the function you comment, including the remark about needing to rewrite?

[mario-bucev]

Good idea! I've added a comment about these issues as requested