Fix duplicate column names returned by plot_smooth_compare()

[cimentadaj]

Summary

• plot_smooth_compare() renames the helper columns age_mid and plot_y to user-facing labels (Age, Exposures, etc.). Because the original tibbles already contain Age/variable, the helper now returns duplicate
  column names.
• Every consumer of smooth_flexible() inherits those duplicates.

Proposed change

./R/graduation.R

  -  names(data_in)[names(data_in) == "age_mid"] <- age_text
  -  names(data_out)[names(data_out) == "age_mid"] <- age_text
  -  names(data_in)[names(data_in) == "plot_y"] <- variable_display
  -  names(data_out)[names(data_out) == "plot_y"] <- variable_display

...

  -  sym_variable_display <- sym(variable_display)
  -  sym_age_text <- sym(age_text)
  +  sym_variable_display <- sym("plot_y")
  +  sym_age_text <- sym("age_mid")

...

  -    labs(title = title_text,
  -         x = age_text,
  -         y = variable_display)
...
  +    labs(title = title_text,
  +         x = age_text,
  +         y = variable_display)

• We keep the translated axis labels via labs() but stop renaming the data frame columns.
• Result: data_adjusted / data_original still carry the helper variables, but with unique names (age_mid, plot_y). Plotly and downstream consumers now work without additional sanitising.

You can test with:

  df <- data.frame(Age = c(0, 1, 5), Exposures = c(10, 20, 30))
  res <- ODAPbackend:::plot_smooth_compare(df, df, "Exposures")
  names(res$data_adjusted)
  # before: "Age" "Exposures" ... "Age" "Exposures"
  # after:  "Age" "Exposures" "single" "AgeInt" "age_mid" "age_label" "plot_y"

Everything else (figure output, labels, translations) remains unchanged.

[cimentadaj]

Actually, I'll submit a PR, don't fix this.

[timriffe]

We had patched this by padding the "translation" columns with a space on either side, but any fix will do if this is cleaner