It took me a while to dreg up what was happening with the Nouveau Littré. However, the recto entry now looks again as it should:
Code:
<ar>
<head><k>recto</k></head><def><b>RECTO,</b> N.*m. [<f>ʀɛkto</f>] (mot lat. <i>recto</i>, s.-e. <i>folio recto</i>, sur le feuillet qui est à l'endroit) <f>•</f>*La première page d'un feuillet*; il est opposé à<i>verso </i>. <f>•</f>* Au pl. <i>Des rectos. </i>*<f>♦</f>*<i>Recto verso, </i>des deux côtés. <i>Faire une impression recto verso. </i></def>
</ar>
Apparently, I had rerun the script without filtering entities based on DOCTYPE and the characters substituted for the hexadecimal codepoints ( E.g. '&#AACD') were never tested before.
@GetKey Could you test whether the Nouveau Littré of Nov15th2021 looks well?