Ok, so I found a solution and I post it if someone is in a noobish day and does the same stupid thing than me.
The disk was marked in red as "unconfigured bad", but since I was sure the disk was healthy and wasn't wtirren by anything during the downtime, I've opened it's menu and selected "make unconfigured good". I also had to add the disk back to my array. Strange thing: the "row" and the "slot" numbering were different. At first I added it to the wrong row but the console returned an error and didn't do anything to the array. When I added it to the correct row, everything was normal.
Then I rescanned all the disks and I was prompted by the console who told me that a configuration was found on the new disk. ("configuration" meaning a raid partition, coming from another controller, but in this case coming from the same controler).
The console asked me if I wanted to import the configuration, and even showed me a preview, as I browsed the available configuration I found one which returned a "degraded" state.
Immediately after clicking "go" I got my array in degraded mode.
Finally I've replaced the failing disk (which was very hot) with a new one and now the array is rebuilding.
It seems that it was in fact a coincidence. The disk failed at the same time I've opened the box to add memory.
(+1 for the preview feature of this RAID controler, I wish there were more previews available for the different steps. Usualy toying with unusual RAID problems is always a leap in the dark)