I think, it would be reasonable to check the project feasibility first. You can assume, that every pF of winding capacity on the HV side consumes several 100 VAr reactive power. Or in a slightly different perspective, you'll have most likely difficulties to achieve a transformer self resonant frequency of 10 MHz and more that would be required to transform an asymmetrical waveform at 2 MHz.
I suggest two realistic alternatives:
- go for a flyback transformer operated at a considerably lower frequency
- go for resonant operation at 2 MHz (similar to a Tesla transformer)