Use hypot for BSplineInterpolation, BSplineApprox

[cgarling]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

Replaces the equivalent of sqrt(x^2 + y^2) with hypot(x, y) in BSplineInterpolation and BSplineApprox as hypot can be more accurate in some situations and is only marginally slower (~1 ns slower for Float64 arguments). Two tests broke locally due to very minor float changes -- I updated these tests and all now pass locally.

The methods that take u::AbstractArray have different expressions that look more like sqrt(x^2 + sum(y.^2)) which can't be replaced with hypot calls so I didn't touch those , e.g.,

DataInterpolations.jl/src/interpolation_caches.jl

Lines 905 to 908 in 618ff4b

	for i in 2:n
	s += √((t[i] - t[i - 1])^2 + sum((u[ax_u..., i] - u[ax_u..., i - 1]) .^ 2))
	l[i - 1] = s
	end