Convolution differences between Jax and Scipy

[lucatelli]

There is an unknown issue where jax.scipy.signal.fftconvolve and scipy.signal.fftconvolve yields different results with LMFIT.

• Investigate possible errors
• a simple convolution test with both cases result in the same output, but not when running an optimisation

[lucatelli]

The issue seems not to be related with Jax. This is related to LMFIT version >1.1.0, for an unknown reason that needs further investigation. The minimization is achieved with LMFIT 1.1.0 for different physical situations. But, for the same situations, there are no convergences with LMFIT 1.2.2.