Parameter fitter is not robust

[ma-sadeghi]

fit_circuit_parameters fails at a moderate rate (maybe ~50% of the times). Currently, all our functions take in initial guess as an input argument, so it's not exactly breaking our stuff. But, for the cases where the user doesn't provide it, it relies on our fit_circuit_parameters, so they don't get an answer ~50% of the time.

Some possible solutions:

• Run circuit fitter multiple times and choose the best (fit_circuit_parameters already accepts iters, so it's easy to implement)
• Use a different circuit fitter (EquivalentCircuits.jl has a parameteroptimization, which I haven't tested thoroughly. There's also ImpedanceFitter)
• Write one from scratch (I've already talked to a lead developer at Julia's SciML ecosystem, and they have a package that "seems" to be much faster than our Python solutions, so we can brute-force our way out of a bad guess, i.e., running the optimizer for hundreds of times without worrying too much about the time it takes)

[ma-sadeghi]

We have a brand new circuit fitter (#90), which addresses this issue to some extent, but let's keep this open for a while until we're convinced that the new circuit fitter meets our needs.