Rewrite derivs & smoothing section of vignette following new calc_deriv fitting functionality

[mikeblazanin]

while moving median works well for getting rid of outliers, at low densities it doesn't help at all with the resolution issue (i.e. sudden jumps from 0.001 to 0.002 because that's the limit of resolution of the plate reader)

moving average would help with that. fitting also helps with that (as long as it's done on the post-median data)

[mikeblazanin]

Also maybe modify the analyze vignette where we did subsets for max percap

[mikeblazanin]

Maybe do derivatives first, then talk about if we can't get things smooth enough we can do raw data smoothing, and then at the end talk about re-calculating derivatives of smoothed values

[mikeblazanin]

Or maybe do all the other analyses with example_data_noiseless (including across all the other vignettes), then localize all of the noise discussion into a single vignette.

That way we don't have to keep dealing with noise in all these other areas, and we can just focus on it in the noise vignette.

[mikeblazanin]

done in v0.12.1.9000