Incorrect Multiplication Result when using NumPy Array Instantiation

[Greek64]

I'm getting wrong results when I'm using Fxps instantiated from a NumPy Array and given a custom dtype.
The same Fxp instantiated via different ways, even if seemingly exactly the same, produces correct results.

---

Minimal Example:

import numpy as np
from fxpmath import Fxp

a1 = Fxp(np.array([1.057311]),dtype='Q5.20')
print(f"A1 = {a1[0]}, dtype = {a1[0].dtype}, mem = {a1[0].val}")
a2 = Fxp(1.057311,dtype='Q5.20')
print(f"A2 = {a2}, dtype = {a2.dtype}, mem = {a2.val}")
b = Fxp('7971079',dtype='Q24.20')
print(f"B = {b}, dtype = {b.dtype}")
c = b * a1[0]
print(f"C = {c}, dtype = {c.dtype}")
d = b * a2
print(f"D = {d}, dtype = {d.dtype}")
e = b * Fxp(a1[0],dtype='Q5.20')
print(f"E = {e}, dtype = {e.dtype}")

Output:

A1 = 1.0573101043701172, dtype = fxp-s25/20, mem = 1108670
A2 = 1.0573101043701172, dtype = fxp-s25/20, mem = 1108670
B = 7971079.0, dtype = fxp-s44/20
C = -8349313.630567551, dtype = fxp-s69/40
D = 8427902.36943245, dtype = fxp-s69/40
E = 8427902.36943245, dtype = fxp-s69/40

As can be seen from the output, Fxp A1 produces incorrect multiplication results.
I don't know if it affects different operations besides multiplication and other conditions that have to be met.