Even then, the perfectly good values of non-corrupt memory table may be different from another unit's perfectly good values.
The products we manufacture all use flash memory to store the calibration constants. Of all the memories which I've dumped over the years, no two are alike, ever.
The way we know a memory has been corrupted, is to read the stored CRC and then perform a CRC on the actual memory's contents.
If they differ, then we have a corrupted table.
Some of our newer products actually store the calibration table twice, and perform a CRC on each.
Upon startup, the processor calculates the CRC from the primary table to determine if there is a corruption. If so, then the backup table is employed.