Could an relayer's orbit remain north of the moon? No. Such an orbit would have a center north of earth.

Orbits about the earth have a center within the earth. An inclined relayer orbital plane must intersect earth's plane along a line passing through the center.
A circular relayer orbit would spend half the time north of the moon's orbital plane and half the time south.
When the relayer is north of the orbital plane does it enjoy line of sight with a northern peak? Not necessarily.
Pictured is a northern lunar communications tower. As the satellite moves towards the moon's equatorial plane, it can fall into a moon shadow.

From the tower's point of view, something within the moon shadow would be below the horizon and out of line of sight.
For looking at eclipses, line of sight and similar problems I like to think of Dandelin Spheres. Borrowing from Dandelin Sphere models, the moon shadow falls within a tangent cone to the spherical moon.
Most orbits having the same periods as the moon would be quickly destabilized by the moon's influence. Exceptions are the Lagrange regions. EML4 and 5 are fairly stable.
If a EML4 or 5 probe where crossing through the moon's equatorial plane, how tall would a polar tower need to be enjoying line of sight?

EML4 distance from moon's center is 1 lunar distance, 384,400 km. Moon's radius is 1738 km
sin(angle)=opposite leg/hypotenuse. So the angle pictured is asin(1738/384400) or about .2º.
An EML4 relayer sitting on the moon's equatorial plane will be able to see surface points to within about .2º of the moon's north pole.
Of course a mountain would be visible even if closer to the pole. Let's imagine a mountain right on the pole. How tall would it have to be to be visible to the relayer?
1738.009 km - 1738 km is about 9 meters. So if your mountain is more than 9 meters high, it's top would be visible to the EML4 relayer sitting on the moon's equatorial plane.
Earth's trojan 2010 TK7 is inclined about 21 degrees to earth's orbital plane. So I would think it's possible for a lunar trojan inclined to the lunar orbit.
The angle between the moon's orbital plane and equatorial plane (a.k.a. obliquity) is 6.58º. See Orbit of the Moon. I am imagining a lunar trojan orbit inclined 6.58º with regard to the moon's orbital plane so that the relayer's plane is parallel to the moon's orbital plane.

I enlarged the moon to try to make the moon's equatorial plane more visible. When the moon passes through the line where the two planes intersect, they will be one and the same plane. At this point the relayer would be able to see a 9 meter tower on the north pole as well as a 9 meter tower on the south pole.
All other times the relayer would be either north or south of the moon's equatorial plane. So polar towers would be visible a little more than half the time.
It seems to me there's two possible trojan orbits that see the towers more than half the time: An EML4 orbit parallel to the moon's equatorial plane and an EML5 orbit also parallel to the moon's equatorial plane. But these two relayers would only be 120º apart. At first glance it seems to me they'd need to be 180º apart to provide a given pole with constant line of sight.
So at this time I can't imagine a pair trojan relayers that could provide constant line of sight to the poles.
Halo orbits about L1 and L2 are also possibilities. But these orbits are a lot closer than L4 and L5 and thus more vulnerable to falling into the moon's shadow as they approach the moon's equatorial plane. Thus less than half of their orbits would enjoy line of sight.
It seems to me 3 is the minimum number in any case.