Fix powerspectrum bands on notebooks

[gabrevaya]

This PR fixes the repeated generation of powerspectrum bands and labels in Jupyter notebooks and Literate caused by those notebooks making Makie compute the final limits twice. With this change, the bands and labels are created only the first time, and the following times only their bounds/positions are modified.

I also reduced the threshold for checking if the solution was saved at regular time steps to avoid spurious warnings.

[harisorgn]

The CI Basics errors have to do with this line

Neuroblox.jl/src/Neurographs.jl

Line 184 in 69b1022

param_pairs = make_unique_param_pairs(params_affect)

where

Neuroblox.jl/src/Neurographs.jl

Lines 84 to 102 in 69b1022

	function make_unique_param_pairs(params)
	# HACK : MTK will complain if the parameter vector passed to a functional affect
	# contains non-unique parameters. Here we sometimes need to pass duplicate parameters that
	# affect states in the loop in LIF_spike_affect! .
	# Passing parameters with Symbol aliases bypasses this issue and allows for duplicates.
	param_pairs = if unique(params) == length(params)
	[p => Symbol(p) for p in params]
	else
	map(params) do p
	if count(pi -> Symbol(pi) == Symbol(p), params) > 1
	p => Symbol(p, "_$(rand(1:1000))")
	else
	p => Symbol(p)
	end
	end
	end
	return param_pairs
	end

I don't see how this is affected by this PR. I tried updating my local environment in case it was an upstream issue but the same tests pass for me.

[harisorgn]

Ok as suspected

Neuroblox.jl/src/Neurographs.jl

Line 94 in 69b1022

p => Symbol(p, "_$(rand(1:1000))")

gave the same number twice. Passes after rerunning.

We could write so that it's impossible to hit duplicates but this is a quick (lazy) fix for now.

[gabrevaya]

Great, all test passed! Should we merge it already?