Je viens de voir ton message, je t'ai envoyé un mail.

I've just had a look at the errata list, and yes, it's not short...

Are these bugs documented somewhere?

"In AES decryption, the round keys are used in reverse order compared to encryption. The first round key for decryption is actually the last round key of the encryption key schedule." 🤦
I was looking for my key, not that....
Now I start to see good key guess!

I made some progress. I couldn't find any correlation. Turned out my traces were not properly aligned. As suggested by
@Baldanos
I made a CPA on the cyphertext just to be sure I can detect something. And yes, this is my input bytes leaking: