Fix px.imshow ignoring aspect='auto' for RGB images - #5788

Link to issue

Closes #5172

Description of change

plotly.js anchors the y axis of an image trace to its x axis by default, so pixels render square (the layout.yaxis.scaleanchor reference describes this default, and scaleanchor: false turns it off). px.imshow sets axis constraints only for aspect="equal", so for RGB and RGBA images, which use go.Image, aspect="auto" changed nothing and the image kept its original aspect ratio.

This PR sets scaleanchor=False on every y axis when px.imshow builds an image trace with aspect="auto". fig.update_yaxes covers each facet's y axis, not only the first one. Heatmap traces get no default anchor, so the 2D path is unchanged. aspect="equal", and the numpy default, still render square pixels.

This also changes xarray RGB images with no aspect argument. Their documented default is "auto", so they now stretch to fill the axes like xarray 2D images already do.

Demo

A 100x100 red RGB image rendered with kaleido at width=800, height=400 and no margins:

• main (d586d22), aspect="auto": the image spans 400x400 px
• this PR, aspect="auto": the image spans 800x400 px
• this PR, aspect="equal": the image spans 400x400 px, unchanged

Testing strategy

• test_image_trace_aspect builds a two-facet RGB figure, with and without binary_string. With aspect="auto" every y axis has scaleanchor == False. With the default aspect every y axis keeps scaleanchor unset. It fails on main (2 failed) and passes here.
• tests/test_optional/test_px: 1227 passed, 10 skipped, 4 xfailed.
• ruff format --check and ruff check pass on the changed files.

AI disclosure

An AI agent (Claude Code) reproduced the issue with a kaleido render, wrote the failing test and the fix, and ran the commands above.

Guidelines

• I have reviewed the pull request guidelines and the Code of Conduct and confirm that this PR follows them.
• I have added an entry to the changelog if needed (not required for documentation PRs).

_{🤖 Written and posted by an AI agent (Claude Code) on behalf of @jayzhou2309.}