5.1.46 has this change:
Performance: While looking for the shortest index for a covering index scan, the optimizer did not consider the full row length for a clustered primary key, as in InnoDB. Secondary covering indexes will now be preferred, making full table scans less likely.
In other words, if you have covering index on * (which is quite common on m:n mapping tables), use it rather than PK. As I have spent my time getting indexing right and having PKs be based on primary access pattern and SKs on secondary access pattern, I hereby not welcome the new change that suddenly reverses the behavior in late GA version.
Not good, when mysqldump queries end up taking 6 days instead of previous half an hour, not good at all.
Update: Oh, MariaDB has this reverted, from their change log:
mybug:39653: reverted as invalid
If only upstream MySQL would take note ;-)