Add seaborn.objects column to the plotting comparison

[tdhopper]

TL;DR

seaborn.objects (import seaborn.objects as so) joins the comparison as its own column: 11 of the 13 plots, bringing the total to 90 example cells. Two plots are deliberately absent — the interface has no confidence-band stat for scatter-with-regression and no loess smoother for stacked-smooth-line-and-scatter — because a lookalike without the CI band or true smoothing would misrepresent the comparison.

Stacked on #24 (add-lets-plot).

Files to review (5, notebook-dominated):

File	Why
Examples.ipynb (start here)	11 new cells tagged package:seaborn-objects plus setup-cell config.
render.py	One line: column ordered right after classic seaborn.
tests/test_plots.py	seaborn-objects added to exactly the 11 implemented slugs.
INTRO.md	Paragraph introducing the interface.

Reviewer notes

• No export machinery needed: so.Plot emits PNG natively in Jupyter. The setup cell sets so.Plot.config.display["scaling"] = 1 (the 0.85 default shrinks output) and pushes the notebook's shared figsize/font rcParams into so.Plot.config.theme, landing at the same ~750 logical px as the other columns.
• Two cells carry site-visible caveats: bar-counts (no coord_flip, so manufacturer maps straight to y) and stacked-kde (.limit() clips the axis but, unlike ggplot2's xlim(), doesn't drop rows before density estimation; gridsize=500 keeps the visible window smooth).
• Setup-cell diff looks bigger than it is: the cell's source changed from a JSON string to the list-of-lines form the other 99 cells use — valid nbformat, no content implication.

Verified: notebook executes clean, 90/90 example cells emit PNGs, render.py succeeds, pytest passes.