Should `np.median` be replaced with `np.mean` for plotting 'Mean code coverage growth over time'?

[leonasdev]

I noticed that when plotting the 'Mean code coverage growth over time' graph, np.median is used as the estimator (L166):

fuzzbench/analysis/plotting.py

Lines 159 to 171 in 2920e74

	axes = sns.lineplot(
	y=column_of_interest,
	x='time',
	hue='fuzzer',
	hue_order=fuzzer_order,
	data=benchmark_df[benchmark_df.time <= snapshot_time],
	ci=None if bugs or self._quick else 95,
	estimator=np.median,
	palette=self._fuzzer_colors,
	style='fuzzer',
	dashes=False,
	markers=self._fuzzer_markers,
	ax=axes)

Shouldn't np.mean be used instead? For example, the order of this graph uses benchmark_rank_by_mean:

fuzzbench/analysis/plotting.py

Line 156 in 2920e74

fuzzer_order = data_utils.benchmark_rank_by_mean(

Is there a specific reason for using np.median here, or am I missing something?

Looking forward to your insights. Thanks.