Add hvPlot (Bokeh) column to the plotting comparison

[tdhopper]

TL;DR

hvPlot's Bokeh output joins the comparison: 11 of the 13 plots, bringing the site to 101 example cells across ten columns. The two gaps are honest ones — hvPlot has no regression-with-CI or loess smoother, so those slugs would misrepresent the library.

Stacked on #25 (add-seaborn-objects).

Files to review (7, notebook-dominated):

File	Why
Examples.ipynb (start here)	11 cells tagged package:hvplot; setup cell adds the Bokeh→PNG bridge described below.
pyproject.toml, uv.lock	hvplot and selenium (runtime-only, for Bokeh export).
render.py	Column hvplot (Bokeh) after plotly.
tests/test_plots.py	hvplot on exactly the 11 implemented slugs.
INTRO.md, CLAUDE.md	Bokeh paragraph expanded; export path documented.

Reviewer notes

• Bokeh still has no headless renderer (a Playwright backend lands in unreleased 3.10), so export goes through Selenium — but with no new CI steps. The setup cell points selenium.webdriver.Chrome at the Chrome for Testing build Kaleido already installs (found via choreographer), Selenium Manager fetches a matching chromedriver, and bokeh.io.export.get_screenshot_as_png runs against that. Falls back to system Chrome, which ubuntu-latest ships.
• Cells stay pure plotting code: the driver is wired into an IPython image/png formatter for HoloViews objects, same pattern as the Lets-Plot column.
• Styling: toolbars stripped, fontscale=1.6, single-pane figures normalized to 750×750; faceted cells size per-panel. stacked-kde and scatter-plot-with-facets carry site-visible caveat notes.
• Focus area: the first CI run of this branch is the real test of the Selenium path on a clean runner — locally it's verified end to end (101/101 cells emit PNGs, render + pytest pass).