You can use whatever current transformer you like with the emonTx/emonPi, provided that the VA rating is enough to generate 1.1 V rms at the maximum current you want to measure, and you’re prepared to change the burden (a resistor on the p.c.b) if necessary. So you can use a c.t. to whatever accuracy you are prepared to afford.
However, it is inherent in the theory of current transformers that they must become less accurate at low currents. Therefore, you should always use the c.t. with a rating that is as close as possible to but greater than the maximum expected current.
But also bear in mind that (to the best of my knowledge), the Owl only guesses power based on a constant voltage (which you should set) and a presumed power factor. The emonPi calculates real power based on voltage and current samples.
For what it’s worth, I’m presently testing emonLibCM. After carefully calibrating, it’s agreeing with the pulse count from my non-smart meter to within 0.04% overall, but there is a detectable difference depending on the current itself - due to the phase errors of the YHDC c.t. being current-dependent. It’s those errors which become smaller, the more money you spend on the c.t. And don’t forget the v.t. either - that too has voltage-dependent phase errors that contribute to the overall error.