InverseTrigonometryAtInfinity​(AngouriMath.​Entity,​AngouriMath.​Entity)

Neither arcsine nor arccosine is real past 1, and this library reads both on the side
of the cut below the real axis: arcsin(t) is pi/2 - i*arcosh(t) for real t greater
than 1 and -pi/2 - i*arcosh(t) for t less than -1, with arccos being pi/2 - arcsin
throughout. arcosh grows without bound, so what is left in the limit is that real part
with an infinite imaginary one -- https://github.com/asc-community/AngouriMath/issues/333.
C99 and Python take the other side of both cuts and so answer the conjugates.
CompiledArcsinBranchTest pins the side this library takes, and a limit that disagreed
with the function it is a limit of would be worse than either convention.
An arcsecant is an arccosine of the reciprocal, so a diverging argument takes it to
arccos(0), which is a right angle and is real. Substituting the infinity does settle
that one, but it settles it as arcsec(+oo) -- the right angle written as a function of
an infinity rather than as the right angle.