Canon already has it by firmware, but resulting correction is written in the file as metadata.
Thus, only the Canon DPP software uses that data to eliminate hot or dead pixels (or even dust).
Many of us use Adobe ACR or other software to develop, and thus this function is useless and cannot automatically correct dead pixels.
It would be great if ML could use the dead pixel information and write the RAW file with the dead pixels corrected, interpolated from their neighbour pixels.