Many PCR manufacturers recommend preparing the reaction mix for the reaction on ice. For example, here is NEB's recommendation for their Q5 polymerase. Other manufacturers also include similar steps:
We recommend assembling all reaction components on ice and quickly transferring the reactions to a thermocycler preheated to the denaturation temperature (98°C).
However, I can't think of a good rationale for this. This ResearchGate thread describes a few reasons, but they don't seem to make much sense.
Primer mispriming onto bacterial gDNA - Highly unlikely, since the initial melting step would melt away these misprimed primers and allow them to prime the correct sites.
Polymerase is proofreading (3'-5' exonuclease) - Proofreading activity should not affect any of the components, since the primer oligos are present in very large quantities, and the template should not be significantly affected by any proofreading, since they already have the correct sequence.
RNA is degraded by the polymerase - An extremely ridiculous answer, since PCR does not use RNA as a template, and RT-qPCR uses reverse-transcribed cDNA as the template. Whether the RNA is stable or not has no bearing on the activity of a DNA-only PCR once the RNA has been reverse-transcribed.
Polymerase is thermolabile - This is probably the most ridiculous answer to the question. The whole point of using polymerases like Taq, Pfu and Q5 is that they are from thermophile organisms, and they are highly thermostable to withstand the high temperatures during PCR.
What, then, is the reason PCR reactions are recommended to be assembled on ice?