Here is an example of something that I think is currently impossible: say you have a dataset like,

df = pd.DataFrame(itertools.product(np.arange(.2, 1.2, .01), np.arange(0, 50)), columns=['alpha', 'x'])
df['alpha'] = df['alpha'].round(2)
df['x_pow_a'] = np.power(df['x'], df['alpha'])

>>> df
      alpha   x     x_pow_a
0      0.20   0    0.000000
1      0.20   1    1.000000
...     ...  ..         ...
4998   1.19  48  100.156023
4999   1.19  49  102.643945

[5000 rows x 3 columns]

and want to make two line plots: one of (x, x_pow_a) and one of (alpha, x_pow_a) cross-filtered each with a Nearest selector that maps to some global filter call it $filt. This so far is possible. But then if you want to use the selector to make a visual representation of what is being selected (like a "ruleX" element, as in the "Stocks" Mosaic example) it becomes impossible because,

(1) You can't have two Nearest selectors on the same plot because they overwrite some event listener.
(2) You can't have a Nearest selector simultaneously update two selections (one local to the chart to draw the rule and one global for the cross-filter).

A visual representation of the plot described in words:

My suggestion is to change the selection member of interactors to optionally be an array. I have implemented a proof of concept for Nearest and it seems to work well. Wanted to get people's thoughts on this before I took it any further.