do not set maxk in tests

[beckermr]

It turns out this is not needed either for the simpler profiles we use.

[ismael-mendoza]

thanks! just to make sure I'm following: AFAIK the metacal tests are the only place we were setting maxk explicitly. Are you suggesting that perhaps if we used a more complicated profile for this test (we currently use exponential convolved with Gaussian PSF), the current maxk implementation for interpolated images will not match galsim as well? Have you found any examples of this?

[beckermr]

Yes. I have not been able to produce an example of this however. I could build a contrived one.