This question is aimed to TRUE Elec Engineers, not fuse or wire directionality believers.

@rodman99999
Without a load it will appear that the cap is passing a voltage in your example. The voltmeter, which is typically 10 megohms, is high enough impedance that it will appear as if the cap is passing voltage, but if there were any significant load you would not see anything like this.

In fact a non-polar will block DC quite effectively- and protect a tweeter from DC coming from an amplifier. The problem with non-polar caps is that they are actually two electrolytics in series. As a result, they make distortion in both directions since electrolytics can draw considerable current if reverse biased.
Now if you took two electrolytics and put them in series ('back to back', so to speak, which is how a non-polar is built) but then **also** biased their junction with a DC voltage, as long as the AC waveform going through them has less amplitude then the DC voltage (thus keeping them forward biased) they can be as low distortion as a good film cap.

Electrolytics have a bad reputation as coupling caps but they can work quite well if one understands how they work.

Less than one minute ago, I concluded an experiment. In series, I connected a .1uF Jupiter, copper foil/paper/wax capacitor, a Quart, 8 Ohm tweeter and the negative, of my Fluke multimeter(set to 2 VDC). Touching the multimeter’s positive lead to the + end of a 1.5V, C cell battery and the capacitor’s open lead, to the battery’s - end, results in a momentary(1-2 sec), 1.5VDC reading on my multimeter. Reversing the polarity, results in the same, every time. If there is full DC getting through, long enough for my multimeter’s digital display to recognize and register it, It’s difficult for me to believe, that(perhaps) 80VDC, from a toasted SS amp’s outputs, wouldn’t fry a tweeter. Color me skeptical.

Interesting followup- I just tried the same experiment, but substituted a ClarityCap ESA, 33uF, 250V cap and got a constant 1.45VDC reading, with either polarity. 

@Rodman99999, I'm not sure how to explain those results, but if I understand correctly the meter was indicating the voltage across a series combination of the battery, the cap, and the tweeter.  Wouldn't it be more meaningful to put the meter across just the tweeter, and put the battery across the series combination of the cap and the tweeter?

Regards,
-- Al 

Same results, for a moment. BUT, now the tweeter’s blown(it literally/audibly, "popped"). Trying to read, across the tweeter’s terminals, wasn’t any more, "meaningful"(since I already knew DC voltage would pass), BUT- decidedly more EXPENSIVE! Explaining the results is easy(ie: non-polarized film capacitors, WILL pass DC).