The smartd process on my server is reporting "1 Currently unreadable (pending) sectors" and the kernel has reported an "Unrecoverable read error".

I found this on raid.wiki.kernel.org:
echo repair > /sys/block/mdX/md/sync_action
which, if I read the page correctly will cause the array to check for inconsistencies and perform a write to correct them.

I can perform the repair sync_action, but I know which sector is bad, is there a way to give this sector to the raid subsystem and have it "repair" it without having to scan the entire 4T array?

if you try writing to the bad block, then it should get reassigned.
use seek option.

Exactly, I want to know how to instruct the raid subsystem to find the correct data that belongs in the bad sector and write that data to that sector so that the bad sector gets remapped and the integrity of the raid array is maintained.