update legend, xlabel & format of matplotlib plots

[wang-boyu]

• change matplotlib plot format from "png" to "svg"
• automatically set legend location through ax.legend(loc="best")
• add text "Step" to x axis label

Before:

After:

[github-actions]

Performance benchmarks:

Model	Size	Init time [95% CI]	Run time [95% CI]
BoltzmannWealth	small	🔵 +0.4% [-0.4%, +1.2%]	🔵 +0.3% [+0.2%, +0.5%]
BoltzmannWealth	large	🔵 -0.9% [-1.4%, -0.5%]	🔵 -1.7% [-2.8%, -0.7%]
Schelling	small	🔵 -0.2% [-0.4%, +0.1%]	🔵 +0.2% [-0.1%, +0.4%]
Schelling	large	🔵 +0.1% [-0.7%, +1.0%]	🔵 +1.6% [-0.5%, +3.9%]
WolfSheep	small	🔵 -0.8% [-1.1%, -0.6%]	🔵 -0.3% [-0.5%, -0.0%]
WolfSheep	large	🔵 -2.2% [-3.2%, -1.0%]	🔵 -4.4% [-6.0%, -2.6%]
BoidFlockers	small	🔵 -2.7% [-3.1%, -2.3%]	🔵 -1.5% [-2.2%, -0.8%]
BoidFlockers	large	🔵 -3.3% [-3.9%, -2.6%]	🔵 -1.5% [-2.3%, -0.8%]

[EwoutH]

All obviously good changes!

One question, are there any implications for performance? Is exporting to SVG slower (or faster) than to PNG?

Otherwise, go ahead and merge.

[rht]

Using png is intentional: #1806 (comment). At the very least, a comment should be added to say it is for performance reason.

[wang-boyu]

Yes svg is indeed very low as compare to png: solara-perf.ipynb (863 ms for png vs. 3.89 s for svg, for a 200x200 schelling model). I also tried increasing png dpi from default 100 to 300, and it becomes slower with no noticeable improvement in terms of plot quality. I have changed back to png format in this PR.