Except that you are in ignorance of the facts. If you are playing back a 250 kHz sine wave with a 2 MHz sample rate, what is the rate of the samples in your buffer? If they are at 2 MHz you will get very good reproduction without *any* resampling at all. If they are much slower than 1 MHz you will start to have problems with your resampling scheme and of course below 500 kHz sample rate you are below the Nyquist rate and the info will be aliased.
Is this really about playing a 250 kHz tone without resampling?