Base.MPFR: don't ccall for special value predicates

MPFR determines special values according to sentinel values of the
exponent field. Although these constants are not documented (they're
defined in MPFR's `src/mpfr-impl.h`), they're now already used in
`Base.MPFR` for converting IEEE 754 to `BigFloat`, so I guess it makes
sense to avoid the `ccall` overhead for predicates like `iszero` and
`isnan`, too.

The context here is that I'm working on generic IEEE 754-`BigFloat`
conversion implementations that would work without using MPFR and
improve correctness (JuliaLang#50642) and performance, so this PR seems like the
obvious prerequisite for being able to use the Julian predicates like
`iszero` without calling libmpfr.

base/mpfr.jl

@@ -220,6 +220,12 @@ for (fJ, fC) in ((:si,:Clong), (:ui,:Culong))
  end
 end
 
+# Warning: the constants are MPFR implementation details from
+# `src/mpfr-impl.h`, search for `MPFR_EXP_ZERO`.
+mpfr_special_exponent(::Val{:zero}) = typemin(Clong) + true
+mpfr_special_exponent(::Val{:nan}) = mpfr_special_exponent(Val(:zero)) + true
+mpfr_special_exponent(::Val{:inf}) = mpfr_special_exponent(Val(:nan)) + true
+
 function BigFloat(x::Float64, r::MPFRRoundingMode=ROUNDING_MODE[]; precision::Integer=DEFAULT_PRECISION[])
  z = BigFloat(;precision)
  # punt on the hard case where we might have to deal with rounding
@@ -234,11 +240,11 @@ function BigFloat(x::Float64, r::MPFRRoundingMode=ROUNDING_MODE[]; precision::In
  z.sign = 1-2*signbit(x)
  if iszero(x) || !isfinite(x)
  if isinf(x)
- z.exp = Clong(2) - typemax(Clong)
+ z.exp = mpfr_special_exponent(Val(:inf))
  elseif isnan(x)
- z.exp = Clong(1) - typemax(Clong)
+ z.exp = mpfr_special_exponent(Val(:nan))
  else
- z.exp = - typemax(Clong)
+ z.exp = mpfr_special_exponent(Val(:zero))
  end
  return z
  end
@@ -976,16 +982,16 @@ for (f,R) in ((:roundeven, :Nearest),
 end
 
 function isinf(x::BigFloat)
- return ccall((:mpfr_inf_p, libmpfr), Int32, (Ref{BigFloat},), x) != 0
+ return x.exp == mpfr_special_exponent(Val(:inf))
 end
 
 function isnan(x::BigFloat)
- return ccall((:mpfr_nan_p, libmpfr), Int32, (Ref{BigFloat},), x) != 0
+ return x.exp == mpfr_special_exponent(Val(:nan))
 end
 
 isfinite(x::BigFloat) = !isinf(x) && !isnan(x)
 
-iszero(x::BigFloat) = x == Clong(0)
+iszero(x::BigFloat) = x.exp == mpfr_special_exponent(Val(:zero))
 isone(x::BigFloat) = x == Clong(1)
 
 @eval typemax(::Type{BigFloat}) = $(BigFloat(Inf))