yep, in infomem_a, technically in the TLV database
however the location of the data never seems to change so TI is comfortable putting an absolute address in the header
but, look further below
"Calibration Data in Info Mem"
TAG_DCO_30 is the TLV header "tag" indicating "this is the DCO30 Calibration Data"
if you were to take the start address of segment Info_A and start "walking" it as a TLV database, eventually you'd find the TLV segment with TAG_DCO_30 as its identifier and you'd know then that you're inside the DCO30 calibration segment
(as it turns out, that would point you around 0x10F8 or so)
once inside there, you can figure out what you want by looking at the address offset from the beginning of that TLV database segment, using the symbolic offsets CAL_DCO_16MHZ, CAL_BC1_16MHZ, etc defined below (notice how they start at 0x0000 and just go up to 0x0007 ... they're "offsets" from the "beginning of the DCO30 TLV database segment")
that would be the "most correct" and "most resilient" way to obtain the calibration values
but, using CALDCO_16MHZ, CALBC1_16MHZ is a shortcut that TI evidently believes works well enough
so long as they never alter the methodologies they use for writing and producing that calibration info on future revisions of the chip... or maybe you find a one-off sample of a chip that had to be recalibrated and it didn't store things at the exact address
but did follow the TLV tag convention