This is a very impressive datasheet. The typical maximum output would be over 3000lm, from a 3.87mm^2 round die (actually 4.2mm^2–the datasheet is in error, see correction below). At 8A (the specified maximum of SFT40), datasheet-typical output is up from ~2100lm to ~2700lm, both due to increased efficacy at the low binning current of 1.5A, and better power handling, i.e., slower leveling out of the output/current curve.
This test already shows the SFT40 reaching 3000lm at 11A. (The tested sample seems to overperform the typical datasheet sample, with 2500lm @8A.) It would not be surprising, then, if a well-binned sample of the SFT42R tests over 4000lm at 11A, with surface intensity exceeding 1000lm/mm^2.
Strange that Luminus is reluctant to publish the datasheet, yet the emitter has been floating around for a while on KD.