1/# conversion is wrong when the number is "close" to the next integer

[Collineli]

To be more exact, the conversion seems to fail at exactly (n-1)/n + 1/(2n) to 1/n, for example for n = 2: 1/2 + 1/4 = 0.75, i.e. the exact value that is in-between 1/2 and 1:

> 1/2 + 1/4 - 0.0001 to 1/2
  (1 / 2) + (1 / 4) − 0.0001 ≈ 1/2            (valid)

> 1/2 + 1/4 to 1/2
  (1 / 2) + (1 / 4) ≈ 0                       (invalid)

Also failing at other N values and when there is whole number added:

> 5 + 6/7 + 1/14 -0.0001 to 1/7

  5 + (6 / 7) + (1 / 14) − 0.0001 ≈ 5 + 6/7   (valid)

> 5 + 6/7 + 1/14 to 1/7

  5 + (6 / 7) + (1 / 14) ≈ 5                  (invalid)

qalc version is 5.4.0, all other 1/# conversions seem to work properly.

[hanna-kn]

Fixed.