It depends on how you can reinforce it. Small sizes require little amounts of steel that might not be practical to manufacture in the laboratory. other limitation is your testing rig, i.e. the capacity of the jacks, the strong floor or the frame to support it etc. However there is no rule of a thumb, but a numerical simulation will help in selecting a practical size. You can use any finite element plate program o try different sizes of slabs with any practical loading to check on the reinforcement required. Then you can fix the size of the slab after meeting the above mensioned constraints.
It is obvious that the space is very small...but I have no other option regarding this. Because the available space offered by the testing machine in not more than 20"X20". Within this size reinforcement at a center to center spacing of 4" to 6" (100mm-150mm) can be provided having a number of 4 to 3. Can this space and number of reinforcement provide proper testing result?
I call it a super-low scale test which is not reliable at all, especially for RC elements. Note that the problem is not only about reinforcement size and spacing, care should be also paid on aggregate sizes. For a 20" in 20" slab (smaller than my TV...!), aggregates might have a major effect on the obtained results.