The GM2 that will be used for the shaking table will be an accelerogram scaled to match the spectrum for a specific period range (i.e. 0.05-0.3s) or scaled to match the entire area of the spectrum? Will it be a recorded or an artificial ground motion?

The GM2 response spectrum should be reasonably close to the ASCE specified spectrum in some period range (undisclosed), while it typically shows lower response intensity in the other period ranges. Remember, the ASCE specified response spectrum is an average one for a large number of ground motions. Therefore spectrum for a specific GM can only be qualitatively checked against the code specified spectrum.