Create Measurements package extension

[chriswaudby]

This PR creates a simple extension for the Measurements package, allowing data with uncertainties to be passed directly for fitting, e.g.:

using LsqFit, Measurements
x = [1.0, 2.0, 3.0]
y = [1 ± 0.1, 2.2 ± 0.2, 2.9 ± 0.3]
model(x, p) = p[1] * x
p0 = [1.]
curve_fit(model, x, y, p0)

The extension creates a simple wrapper for LsqFit.curve_fit that dispatches on ydata::AbstractArray{Measurement{T}} where T. Values and uncertainties are extracted from the input, then weights are calculated as the inverse square of the uncertainties and passed to the main curve_fit function.

A unit test checks that the extension gives the same results as when weights are passed manually.

An __init__ function is defined in LsqFit.jl to load the extension for Julia versions <1.9 (see discussion at https://discourse.julialang.org/t/package-extensions-for-julia-1-9/93397).