Fix px error bars ignoring opacity on categorical traces

[00200200]

Link to issue

Closes #4353

Description of change

Plotly.js error bars have a color property but no opacity, so px.scatter(..., color=..., opacity=0.6, error_y=...) faded the markers and left the error bars fully opaque.

For categorically colored traces (one marker color per trace), this now copies that color onto error_x / error_y / error_z as an rgba(...) string that includes the requested opacity.

Continuous color still cannot follow per-point marker colors: error_y.color is a single color in the schema, not an array. That path is unchanged and covered so opacity no longer has to special-case it.

Demo

import plotly.express as px

df = px.data.iris()
df["e_plus"] = df["sepal_width"] / 100
df["e_minus"] = df["sepal_width"] / 40
fig = px.scatter(
    df, x="sepal_width", y="sepal_length", color="species",
    error_y="e_plus", error_y_minus="e_minus", opacity=0.6,
)
print(fig.data[0].marker.opacity, fig.data[0].error_y.color)
# 0.6 rgba(99, 110, 250, 0.6)

Testing strategy

Added unit tests in tests/test_optional/test_px/test_px.py for categorical error_y + opacity, categorical error_x + opacity, the no-opacity inheritance path, and continuous color + opacity.

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).