The specification does not appear to incorrect but awkward and uncommon..
Datum "C" is not required as an orientation of Datum's A and B is not a required refinement.
I elevate my laser head high and run the program with laser power off and observe the travel.
This is a good idea on the first setup and cut for anything you CNC program.
By default tolerances of location applied to threaded feature apply the axis as derived from the pitch circle (diameter). Since the variation of pitch diameter of a 8-32 unc thread is about 0.0015 (verify) the "Bonus" tolerance at LMC is minimal
Assuming that there is actual logic in this...
"IF" datum A was a datum target line, with a specified location it would work. Or, if an additional datum (B?) was specified in the FCF which would orient Datum A - that would work.
As is - the orientation relative to Datum A is unspecified.
"Can someone explain to me why you would ever desire to basic a print at the MMC with one sided tolerance?"
Old school..
Before CNC's we manufactured stuff with manually operated machines and the strategy for the machine operator (machinist) was to remove just enough material to achieve MMC...
This is correct but only for selected form tolerances applied to the size dimension or FOS. Same applies for straightness specified to the FOS but not the surface exclusively.
Excludes cylindricity...
When a flatness is specified to a surface it is a refinement of the surface, thus the limits...