gh-36165: Fix reduction of cusps for Gamma(N)

    
The method reduce_cusp for Gamma(N) did not work properly on cusps u/v
where (u,v) is congruent to (-1,0) mod N.  This fixes that by fixing the
function _lift_pair().  A doctest has been added with the example
reported on sage-devel.

Fixes #36163

### 📝 Checklist

- [x] The title is concise, informative, and self-explanatory.
- [x] The description explains in detail what this PR is about.
- [x] I have linked a relevant issue or discussion.
- [x] I have created tests covering the changes.
- [x] I have updated the documentation accordingly.
    
URL: #36165
Reported by: John Cremona
Reviewer(s): Frédéric Chapoton

src/sage/modular/arithgroup/congroup_gamma.py

@@ -240,6 +240,12 @@ def reduce_cusp(self, c):
             sage: G = Gamma(50)
             sage: all(c == G.reduce_cusp(c) for c in G.cusps())
             True
+
+        We test that :trac:`36163` is fixed::
+
+            sage: Gamma(7).reduce_cusp(Cusp(6,7))
+            Infinity
+
         """
         N = self.level()
         c = Cusp(c)
@@ -341,7 +347,7 @@ def _lift_pair(U,V,N):
     u = U % N
     v = V % N
     if v == 0:
-        if u == 1:
+        if u == 1 or u == N-1:
             return (1,0)
         else:
             v = N